Coffee roasting

MISSING IMAGE

Material Information

Title:
Coffee roasting color and aroma-active sulfur compounds
Physical Description:
1 online resource (171 p.)
Language:
english
Creator:
Azeredo,Alberto Monteiro
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Food Science and Human Nutrition
Committee Chair:
Marshall, Maurice R
Committee Members:
Rouseff, Russell L
Sims, Charles A
Balaban, Murat O
Welt, Bruce A

Subjects

Subjects / Keywords:
aroma -- coffee -- color -- kinetics -- roasting -- sulfur
Food Science and Human Nutrition -- Dissertations, Academic -- UF
Genre:
Food Science and Human Nutrition thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Roasting is the most important unit operation to develop flavor, aroma and color in coffee technology. Roasting is a highly complex process that combines heat transfer to the beans, moisture loss, and many chemical reactions. The color kinetic parameters of coffee roasted at constant temperatures of 170?C, 180?C, 190?C and 200?C were determined in a lab oven. First and second order kinetics were considered for L* values. Color of ground coffee changed from green (L*=64.08?1.36, a*=-1.34?0.18, b*=14.12?0.43) to dark brown (L*=12.54?0.78, a*=7.63?0.77, b*=8.84?1.23) during the various isothermal roasting processes. The color change reaction rate for L* followed Arrhenius behavior with activation energy of 128.7 KJ?mol-1?K-1 and 145.9 KJ?mol-1?K-1 for first and second order reactions, respectively. Prediction of L* values in whole-bean roasting under non-isothermal conditions demonstrated that second order kinetics produced better agreement with experimental data. Color of the Agtron Roast Color Classification System was assessed using various instruments. Machine Vision under polarized lighting obtained the highest sensitivity. Cross-matching between sensory analysis results and the matching obtained from the instruments demonstrated also that Machine Vision under polarized lighting was capable of matching the color of coffee roasts with the classification disks 60.1% of the time. Analysis of aroma-active sulfur compounds demonstrated that the time-temperature profiles significantly affect their final concentrations. Sulfur-volatiles are known to produce many off-flavors, though pleasant aromas are also caused by many sulfur-containing volatiles when at the right concentrations. Total sulfur-volatiles were significantly higher in moderately dark and dark coffee roasts when a long-time profile was applied, in contrast with a short-time profile for the same final degree of roast. However when individually analyzed, compounds such as 2-furfurylthiol, a major character impact volatile, were produced in significantly higher concentrations in the short-time profile. Findings from this study can help make color prediction possible during roasting in real time. There is evidence that roasting profile greatly affects sulfur-compound formation, but further studies need to be conducted to determine the best sulfur volatile balance that will lead to coffee with better aromas.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Alberto Monteiro Azeredo.
Thesis:
Thesis (Ph.D.)--University of Florida, 2011.
Local:
Adviser: Marshall, Maurice R.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2011
System ID:
UFE0043168:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 COFFEE ROASTING : COLOR AND AROMA ACTIVE SULFUR COMPOUNDS By ALBERTO MONTEIRO CORDEIRO DE AZEREDO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011

PAGE 2

2 2011 Alberto Monteiro Cordeiro de Azeredo

PAGE 3

3 To God for giving me the most caring wife, the most precious daughter, and the most supportive parents and family who helped me in ach ieving this honorable milestone

PAGE 4

4 ACKNOWLEDGMENTS I would like to give my deepest thank s to my advisors Dr. Balaban and Dr. Marshall for giving me this opportunity to pursue my PhD degree and for their support, advice and friendship throughout this experience. Thanks to Dr. Rouseff for his invaluable help, assistance and time for my research, thanks for letting me be part of his lab I al so want to thank my other committee members, Dr. Sims and Dr. Welt, for all their expertise and he lp. I acknowledge all the assistance provided to me from the University of Florida, the Institute of Food and Agricultural Sciences, and the department of F ood Science and Human Nutrition. I would also like to give my special thanks to Dr. Susan Percival f or her continuous encouragement to pursue a PhD. I would like to thank Mr. Washington Rodrigues and Mr. Edgard Bressani from Ipanema Coffees, located in southern Minas Gerais, Brazil, for kindly providing green coffee beans for this research. I would like to thank my lab mates Milena and Maria for all their help and suggestions and would like to extend my special thanks to Dr. Asli Odabasi and Lorenzo Fuentes for their time and expertise Most of all, I thank my wife Gabriella, who gave me the strength to succeed and the support to accomplish my goals. I thank also my mom Raquel, for her invaluable advices and most of all for being an important influence in my life For my daughter Anna, who was born during my PhD program, and gave me extra enthusiasm to keep me going towards my goal. T o my family in Brazil, I would like to give thanks for their encouragement I thank my American parents Shirley and Carl Romey, for their unconditional support and love. All of you made my moments in Gainesville the best t ime of my life.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF ABBREVIATIONS ................................ ................................ ........................... 13 ABSTRACT ................................ ................................ ................................ ................... 15 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 17 Justif ication ................................ ................................ ................................ ............. 17 Objectives of Study ................................ ................................ ................................ 18 Specific Objectives ................................ ................................ ................................ .. 18 2 LITERATURE REVIEW ................................ ................................ .......................... 20 Coffee Origin and Market ................................ ................................ ........................ 20 Green Coffee Post Harvest Processing ................................ ................................ .. 21 Dry or Natural Processing ................................ ................................ ................ 21 Wet Processing ................................ ................................ ................................ 22 Coffee Roastin g ................................ ................................ ................................ ...... 23 General Aspects ................................ ................................ ............................... 23 The Process ................................ ................................ ................................ ..... 24 Heat and Mass Transfer ................................ ................................ ................... 24 Coffee Roasting Methods ................................ ................................ ................. 25 Rotating cylinder ................................ ................................ ........................ 25 Fixed drum with rotating paddles ................................ ............................... 26 Fluidized bed ................................ ................................ .............................. 26 Chemical Compo sition of Green and Roasted Coffees ................................ ........... 27 Carbohydrates ................................ ................................ ................................ .. 27 Lipids ................................ ................................ ................................ ................ 27 Protein ................................ ................................ ................................ .............. 28 Alkaloids ................................ ................................ ................................ ........... 28 Chlorogenic Acids ................................ ................................ ............................ 29 Organic Acids ................................ ................................ ................................ ... 31 Physical Changes in Coffee during Roasting ................................ .......................... 32 Moisture ................................ ................................ ................................ ............ 32 Surface and Structur e of the Bean ................................ ................................ ... 32 Color ................................ ................................ ................................ ................. 33 Coffee Aroma and Flavor Development During Roasting ................................ ....... 34 Formation of Aroma Compounds ................................ ................................ ..... 36 Coffee Aroma Changes During Storage ................................ ........................... 37

PAGE 6

6 3 COFFEE COLOR: KINETICS OF COLOR FORMATION DURING THE ROASTING PROCESS ................................ ................................ ........................... 52 Materials and Methods ................................ ................................ ............................ 54 Sample Preparation ................................ ................................ .......................... 54 Color Analysis ................................ ................................ ................................ .. 55 Kinetic Considerations ................................ ................................ ...................... 56 L* values ................................ ................................ ................................ .... 57 a* and b* values ................................ ................................ ......................... 57 Statistical Analysis ................................ ................................ ............................ 58 Results ................................ ................................ ................................ .................... 58 Thermal Treatments ................................ ................................ ......................... 58 L* values ................................ ................................ ................................ ........... 58 a* and b* values ................................ ................................ ............................... 60 Summary ................................ ................................ ................................ ................ 61 4 COMPARISON OF MINOLTA, HUNTERLAB AND MACHINE VISION IN A SSESSING THE COLOR OF ROASTED COFFEE AND THE AGTRON SCAA ROAST COLOR CLASSIFICATION SYSTEM ................................ ........................ 73 Materials and Methods ................................ ................................ ............................ 75 Agtron SCAA Roast Color Classification System ................................ ............. 75 Coffee and Roaster Equipment ................................ ................................ ........ 76 Color Analysis ................................ ................................ ................................ .. 77 Sensory Analysis ................................ ................................ .............................. 79 Statistical Analysis ................................ ................................ ............................ 80 Results and Discussion ................................ ................................ ........................... 80 Comparison of Color Measurement of the Agtron RCCS with Different Instruments ................................ ................................ ................................ ... 80 Comparison of Color Measurement of Roasted Coffees Using Different Instruments ................................ ................................ ................................ ... 82 Correlation Between Sensory Analysis and Instrumental Measurements ........ 83 Summary ................................ ................................ ................................ ................ 86 5 COFFEE AROMA: FORMATION AROMA ACTIVE SUL F UR VOLATILES DURING COFFEE ROASTING ................................ ................................ .............. 97 Materials and Methods ................................ ................................ .......................... 100 Coffee and Roaster Equipment ................................ ................................ ...... 100 Color Analysis ................................ ................................ ................................ 101 Degree of Roast ................................ ................................ ............................. 102 Sulfur Volatile Analysis ................................ ................................ ................... 103 Statistical Analysis ................................ ................................ .......................... 104 Results and Discussions ................................ ................................ ....................... 105 Coff ee Roast ................................ ................................ ................................ .. 105 GC Identifications and Quantifications ................................ ............................ 105 Principal Component Analysis (PCA) ................................ ............................. 109

PAGE 7

7 Summary ................................ ................................ ................................ .............. 110 6 SUMMARY AND CONCLUSIONS ................................ ................................ ........ 123 APPENDIX A ROASTER AND COFFEE PICTURES ................................ ................................ 125 B SENSORY COLOR MATCHING ................................ ................................ .......... 126 Screening Test ................................ ................................ ................................ ...... 126 Questionaire ................................ ................................ ................................ ......... 131 C SULFUR COMPOUNDS PEAK AREAS ................................ ............................... 136 D TIME TEMPERATURE TREATMENTS AND C OFFEE ROAST COLOR VALUES ................................ ................................ ................................ ................ 140 LIST OF REFERENCES ................................ ................................ ............................. 162 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 171

PAGE 8

8 LIST OF TABLES Table page 2 1 Coffee production for the 6 largest producing countries (in thousands of 60 kg bags). ................................ ................................ ................................ ............. 38 2 2 Green coffee price indicator in February 2011 (in cents of US dollars per kg). ... 38 2 3 Sugar content of green arabica and robusta coffees. ................................ ......... 39 2 4 Typical content of the lipid fraction in green coffee. ................................ ............ 39 2 5 Chlorogenic acid content in green and roasted C. arabica Cv. Bourbon (1), C. arabica Cv. Longberry (2), and C. canephora Cv. robusta (3). ........................... 40 2 6 Acid contents in different green arabica and robusta coffee samples analyzed by HPLC/UV. ................................ ................................ ................................ ...... 40 2 7 Guide used in color roasting with information about the end of roast average temperatures, Agtron color and some of the common descriptors used for coffee roasts. ................................ ................................ ................................ ...... 41 2 8 Number of volatile compounds reported in green and roasted coffee. ............... 42 2 9 Concentrations in medium roasted arabica coffee from Colombia divided into groups of volatile compounds with similar odor qualities. ................................ ... 4 3 2 10 Odor thresholds and Odor Activity Values (OAV) of some important odorants from roasted samples of arabica and robusta coffees. ................................ ....... 44 2 11 Odor Activity Values (OAV) of some important odorants from a Colombian arabica coffee roasted to light (l), medium (m) and dark (d) degree of roast ..... 44 2 12 Loss of potent odorants from ground roasted Colombian coffee at room temperature. ................................ ................................ ................................ ....... 45 3 1 List of treatment temperatures and times. ................................ .......................... 62 3 2 Settings used in the Nikon D200 camera. ................................ .......................... 62 3 3 Coefficients of determination (r 2 ) for the plots of zero, half, first and second order kinetics for L* values. ................................ ................................ ................ 62 3 4 Rate constant values calculated for first and second order reactions of L* values. ................................ ................................ ................................ ................ 62 3 5 Model parameters found by curve fitting for a* and b* values. ........................... 63

PAGE 9

9 3 6 Prediction error calculated in SS, and average difference ( ) for a* and b* values at 170, 180, 190 and 200 C ................................ ................................ ... 63 4 1 Agtron SCAA Roast Color Classification disk numbers and correspondent degree of roast descriptions ................................ ................................ .............. 88 4 2 List of coffee roast numbers and their final roasting temperatures. .................... 88 4 3 Settings used in the Nikon D200 camera. ................................ .......................... 88 4 4 Average L*, a*, and b* values from Machine Vision, Minolta and HunterLab of the Agtron RCCS disks and those of standard red plate, and their color representation. ................................ ................................ ................................ .... 89 4 5 Average L*, a*, and b* values from machine vision, Minolta and HunterLab of green and roasted ground coffee, and their color representation. ...................... 90 4 6 Number of panelists divided by age group in the sensory study. ........................ 91 4 7 Sensory color matching of ground roasted coffee samples with the Agtron RCCS disks. Data represents the number of times each classification disk was selected per coffee roast sample by the panelists. ................................ ...... 91 4 8 ................................ ................................ ................................ .................... 92 5 1 Settings used in the Nikon D200 camera under polarized lighting. ................... 113 5 2 Final sampling temperature, L* values, average L* values, and degree of roast description of coffee roasted under HTLT and LTLT profiles. .................. 113 5 3 List of sulfur volatiles analyzed in this study. ................................ .................... 114 C 1 Data for sulfur compounds found in coffee experiment (Chapter 5). ................ 137 D 1 Time temperature of the roasting treatments used to obtain coffee samples in an Ambex Roaster. ................................ ................................ ........................... 140 D 2 L*, a*, and b* values of c offee samples roasted under non isothermal conditions in an Ambex Roaster at various temperatures. ............................... 161

PAGE 10

10 LIST OF FIGURES Figure page 2 1 Coffee cherry longitudinal section showing the cross section of the beans, silverskin, pulp, parchment and skin. ................................ ................................ .. 46 2 2 Heat and mass transfer aspects during coffee bean roasting. ............................ 46 2 3 Basic schematics in different roasting technologies. ................................ .......... 47 2 4 Trigonelline content in green and roasted coffee beans classified by cup quality (cup quality increases from right to left). Soft, Hard, Rioysh, Rio and Rio Zona are grades given to arabica coffees. ................................ ................... 48 2 5 Structure of chlorogenic acids (CGA) identified in coffee. ................................ 49 2 6 5 CQA/caffeine ratio versus L* values in coffee roasted to different roast degrees. ................................ ................................ ................................ ............. 50 2 7 Development of titratable acidity as a function of organic roast loss for Colombian arabica and Indonesian robusta. ................................ ...................... 50 2 8 Relative bean volume increase as a function of roast loss under HTST and LTLT roasting conditions. (Source: Schenker and others 2000) ......................... 51 3 1 Typical bean time temperature profile in drum coffee roasters. .......................... 64 3 2 Example of image taken from the heat treated ground coffee samples for color analysis. ................................ ................................ ................................ ..... 64 3 3 Time temperature data recorded for samples treated at 170C, 180 C, 190 C and 200 C for different sampling times. ................................ .............................. 65 3 4 Plots for zero, half, first and second order of L* values versus time for samples treated at isothermal conditions at 170, 180, 190 and 200 C. .............. 66 3 5 Arrhenius plot of first and second order reactions for L* values showing the obtained r 2 ................................ ................................ ................................ ......... 67 3 6 Plots of predicted and experimental L* values versus time at isothermal conditions of 170 180, 190 and 200C, for ground green coffee considering first order reactions. ................................ ................................ ............................ 68 3 7 Plot of predicted and experimental L* values versus time at non isothermal conditions, for whole beans roasted i n Ambex drum roaster, considering first order reactions. ................................ ................................ ................................ ... 69

PAGE 11

11 3 8 Plot of predicted and experimental L* values versus time at non is othermal conditions, for whole beans roasted in Ambex drum roaster, considering second order reactions. ................................ ................................ ...................... 69 3 9 Plots of a* and b* values of the samples treated at isothermal conditions at 170 ( ), 180 ( ), 190 ( O ) and 200 C ( + ) versus time. ................................ ......... 70 3 10 Plots of predicted and experimental a* values versus time at isothermal conditions of 170 180, 190 and 200C, for ground green coffee. ....................... 71 3 11 Plots of predicted and experimental b* values versus time at isothermal conditions of 170 180, 190 and 200C, for ground green coffee. ....................... 72 4 1 Roasting profile used to produce roasted coffees in this study. .......................... 93 4 2 RCCS disks assessed using machine vision, HunterLab, and Minolta colorimeter. ................................ ......... 94 4 3 using machine vision, HunterLab, and Minolta colorimeter. ............................... 95 4 4 Cross matching of the color obtained from the sensory analysis and from the ................................ ............................ 96 5 1 Drawings of (1) the original Pasteur glass pipet showing the cutting section, (2) the sample holder and (3) the complete sampling system. ......................... 115 5 2 Roasting profiles of HTST (solid line) and LTLT (dashed line) used to produce the roasted coffees in this study. ................................ ........................ 115 5 3 Superposed chromatograms of two GC runs showing the spiked 2 furfurylthiol. ................................ ................................ ................................ ....... 116 5 4 Evolution of total sulfur compounds in coffee during LTLT (dashed line) and HTST (solid line) roasting profiles ................................ ................................ ..... 116 5 5 Comparison between chromatograms of light coffee roast produced under HTST and LTLT roasting profiles. ................................ ................................ ..... 117 5 6 Comparison between chromatograms of dark coffee roasts produced under HTST and LTLT roasting profiles. ................................ ................................ ..... 117 5 7 Evolution of sulfur compounds for coffee during the LTLT (dashed line) and HTST (solid line) roasting profiles. ................................ ................................ .... 118 5 8 Evolution of sulfur compounds as OAV for coffee during the LTLT (dashed line) and HTST (solid line) roasting profiles. ................................ ..................... 120

PAGE 12

12 5 9 Loading plot showing principal component 1 (PC 1) and principal component 2 (PC 2) for the volatiles studied. ................................ ................................ ..... 121 5 10 PCA score plot showing Principal Component 1 (PC 1) and Principal Component 2 (PC 2) for the volatiles studied. ................................ .................. 122 A 1 Pictures of the drum roaster (A), front section detailing the sampling system (B), cooling of roasted coffee samples (C), ProfilePlusDCQ automation system main screen (D). ................................ ................................ ................... 125

PAGE 13

13 LIST OF ABBREVIATION S 3M2BT 3 methyl 2 butenthiol a* redness / greenness values from the L*, a*, and b* color space DMDS dimethyl disulfide FFT 2 furfurylthiol FMS furfuryl methyl sulfide HLse Hunterlab in specular excluded mode HLsi Hunterlab in specular included mode HTST High temperature short time J Joule k temperature dependent rate constant K temperature in Kelvin k 0 reaction rate constant kg kilogram L* lightness values from the L*, a*, and b* color space L* yellowness / blueness values from the L*, a*, and b* color space LTLT Low temperature long time m meter MET methional Mi Minolta mol mole MVnp Machine vision under non polarized lightin g MVp Machine vision under polarized lighting OAV Odor Activity Value PC1 Principal Component 1

PAGE 14

14 PC2 Principal Component 2 PCA Principal Component Analysis sec second TPN thiophene g microgram

PAGE 15

15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy COFFEE ROASTING : COLOR AND AROMA ACTIVE SULFUR COMPOUNDS By Alberto Monteiro Cordeiro de Azeredo August 2011 Chair: Maurice R. Marshall Major: Food Science and Human Nutrition Roasting is the most important unit operation to develop flavor, aroma and color in coffee technology. Roasting is a highly complex process that combines heat transfer to the beans, moisture loss, and ma ny chemical reactions. The color kinetic parameters of coffee roasted at constant temperatures of 170C, 180C, 190C and 200C were determined in a lab oven. First and second order kinetics were considered for L* values Color of ground coffee changed from green (L*=64.081.36, a*= 1.340.18, b*=14.120.43) to dark brown (L*=12.540.78, a*=7.630.77, b*=8.841.23) during the various isothermal roasting processes. The color change reaction rate for L* followed Arrhenius beh avior with activation energy of 128.7 KJ mol 1 K 1 and 145 .9 KJ mol 1 K 1 for first and second order reactions respectively Prediction of L* values in whole bean roasting under non isothermal conditions demonstrated that second order kinetics produced be tter agreement with experimental data. Color of the Agtron Roast Color Classification System was assessed using various instruments. Machine Vision under polarized lighting obtained the highest sensitivity. Cross matching between sensory analysis results a nd the matching obtained from the

PAGE 16

16 instruments demonstrated also that Machine Vision under polarized lighting was capable of matching the color of coffee roasts with the classification disks 60.1% of the time Analysis of aroma active sulfur compounds demon strated that the time temperature profiles significantly affect their final concentrations. Sulfur volatiles are known to produce many off flavors, though pleasant aromas are also caused by many sulfur containing volatiles when at the right concentrations. Total sulfur volatiles w ere significantly higher in moderately dark and dark coffee roasts when a long time profile was applied in contrast with a short time profile for the same final degree of roast. However when individually analyzed, compounds such a s 2 furfurylthiol, a major character impact volatile, were produced in significantly higher concentrations in the short time profile Findings from this study can help make color prediction possible during roasting in real time. There is evidence that roasting profile greatly affects sulfur compound formation, but further studies need to be conducted to determine the best sulfur volatile balance that will lead to coffee with better aromas.

PAGE 17

17 CHAPTER 1 INTRODUCTION Justification Coffee is a widely consumed beverage prepared from the roasted beans of the coffee plant. The importance of coffee in the world economy cannot be overstated. It is o ne of the most valuable primary products in world trade, in many years second in value only to petroleum as a source of foreign exchange to producing countries ( Pendergrast 2009 ; ICO 2010b ) Excluding water, c offee is the most popular beverage in the US among adults ahead of carbonated drinks, tea and milk products ( Storey and others 2006 ) With an estimated $ 16.6 billion value in 2008, green coffee was the world's sixth largest legal agricultural export, behind wheat, soybeans, palm oil, maize and rice Global output increased from 6.2 million metric tons in 1996 to 8 million tonnes in 20 10, and Brazil stands out as the world leader in green bean production with 2.4 mill ion metric tons ( FAOSTAT 2011 ) The chemical composition of green coffee beans has been extensively studied and differs significantly depending on the specie s cultivation practi ces, origin and processing methods. However most of t he characteristic formation of color, aroma and flavor of the finished product depend on one processing step: the roasting. During this time temperature dependent process major compositional changes tak e place, including loss of water, decrease s in protein, amino acids, arabinogalactan, reducing sugars, and trigonelline, loss of sucrose and chlorogenic acid, and the formation of melanoidins, volatiles and carbon dioxide ( Parliment and Stahl 1995 ) producing the compounds responsible for color, aro ma and flavor of roasted coffee ( Houessou and

PAGE 18

18 others 2007 ) As a key step in coffee technology, roasting is not yet completely understood and is mainly performed on an empirical basis C olor is the main monitor ing parameter used for the degree of roast. Changes in color may happen extremely fast during roasting, and a reliable time temperature monitoring system is not yet existent. Coffee aroma and flavor are also greatly affected by the roasting process. Studie s have shown that roast ing coffee to the same color but under different time temperature conditions may produce coffee with different aroma profiles ( Schenker and others 2002 ; Baggenstoss and others 2008b ) Objectives of Study Understanding color formation and its kinetics during coffee roasting will make it possible to develop real time predictions of the degree of roast, based on the time temperature process. Moreover, roasting coffee to the same color using two dissimilar time temperature profiles and tracing the formation of some aroma active sulfur co mpounds will provide valuable information about optimized aroma at any degree of roast. Specific Objectives 1. To determine the kinetics parameters of color change in coffee roasting by using ground green coffee beans under isothermal conditions at 170C, 180 C, 190C and 200 C, and to develop a model to predict color based on time temperature roasting profiles. 2. To determine the L*, a*, and b* values of the Agtron SCAA Roast Classification System with different instruments and correlate these values with real roasted coffee samples, using instrument ation and sensory analysis.

PAGE 19

19 3. To determine the effect of roasting coffee to the same degree of roast (color), under two dissimilar time temperature profiles on the formation of sulfur aroma active compounds including 2 furfurylthiol, methional, furfuryl methyl sulfide, dimethyl disulfide, thiophene, 3 methyl 2 butenthiol, dimethyl trisulfide, 4 mercapto 4 methyl pentan 2 one and 2 acetyl 2 thiazoline.

PAGE 20

20 CHAPTER 2 LITERATURE REVIEW Coffee Origin and Market The coffee plant from the genus Coffea belongs to the Rubicacea family. Coffe e and its origins can be traced to Yemen and to the highlands of Ethiopia where the first beverage was made from it. The Arabs were the first to cultivate it in the 14th century and f rom there, i t was brought to the new world and to the rest of the sub tropics by the 17th century ( Flament 2001 ; Buffo and Cardelli Freire 2004 ; ICO 2010b ) It is now cultivated throughout the world and is one of the most consumed beverages. The world's largest coffee producing region is Latin America and the Caribbean, and Brazil stands out as the largest producing country ( FAOSTAT 2011 ) Two plant species account for more than 99% of commercial coffee production: C offea arabica L. ( arabica coffee) with more than 70% of the total world production and C. canephora (robusta coffee). A third species, C. liberica has very little contribution to the world coffee market Production of arabica and robusta coffees in the world presents year to year fluctuations due mainly to frost and droughts in the producing regions. In 2010 the world production was over 134 million of 60 kg bags, representing more than 8 million tons of coffee (Table 2 1 ) Within the six largest coffee producers, only Colombia and Ethiopia harvest arabica coffee exclusively The other countries have a m ixed production of bo th arabica and robusta coffees ( ICO 2010a ) Today, coffee cultivation is widespread in tropical and subtropical regions, with the bulk of arabica coffee concentrated in Latin America and robusta coffee in Southeast Asia and Africa ( Anzueto and others 2005 )

PAGE 21

21 Green coffee is the second most traded commodity after petroleu m in value and is one of the most popular beverages in the world ( Parras and others 2007 ) The trade value of coffee varies with the origin and species. Typically a rabicas are at least twice as valuable as robustas. Mild a rabicas from Colombia are highly regarded and present higher trade values than other mild a rabicas (Table 2 2) ( ICO 2010a ) Green Coffee Post Harvest Processing The coffee plant is a small tree of 4 6 and 8 12 meters tall for a rabica and robusta coffees respectively. Fruits mature between seven a nd eleven weeks from the flowering and the green fruits will turn yellow or red, depending on the variety during ripening The fruit, usually called cherry, contains two seeds embedded in a fleshy pericarp and a mucilage layer (Figure 2 1). The seeds, whic h are the green coffee beans, are generally smaller and rounder in robusta coffees ( Anzueto and others 2005 ; Arya and Rao 2007 ) One of the main characteristics of high quality coffee is the good organoleptic properties which depend on the spe cie s and on the post harvesting processing steps involved. To maintain a high quality, special attention needs to be given to a number of stages including sorting and drying of the cherries, and roasting ( Bee and others 2005 ) There are typically three to four processing alternatives for the harvested cherries : the dry and the wet methods, and a few variations of these two techniques. In t h is processing step the chemical composition of the beans can be greatly affected Proper technique and attention to details will minimize the occurrence of potential defective beans. Dry or N atural P rocessing This method consists of drying the coffee cherries as a whole. T he cherries wit h the bean, mucilage and pulp are sun dried on open patios, or dried mechanically. This

PAGE 22

22 process requires that the cherries are harvested only at the over ripe stage or dry stage. This is because ripe cherries will lead to undesirable fermentation, especial ly if the drying is not done the same day of harvesting. T his processing method produces coffee with good body, sweetness, smoothness and complexity. Dry processing is typically used in coffees from Brazil, Ethiopia, India and Ecuador due to the abundance of sun during the harvest season It is also considered an environmentally friendly method due to the very low use of water in contrast with wet processing ( Bee and others 2005 ) W et P rocessing The wet method requires the use of specific equipment and large quantities of water. When properly done, it ensures that the intrinsic qualities of the coffee beans are better preserved, producing a green coffee which is homogeneous and has few defective beans ( Bee and others 2005 ) This method consists of processing the beans by several steps. It is necessary that only ripe cherries are used. Ripe cherries are passed through a small mill, which compresses the cherries without affecting the bea n. The hard texture of the cherry allows spitting the bean when a small pressure is applied. The beans are put into tanks where the mucilage is removed by fermentation. The beans are covered with water and ferment by the naturally occurring bacteria in the beans. This process takes 12 to 36 hours depending on the local temperature and conditions. Great d isadvantage s of this method are the high volume of waste water generated and the risk of formation of over fermented beans However this method produces cleaner, brighter, fruitier and more acidic beans, higher in marke t value ( Arya and Rao 2007 ; Kleinwachter and Selmar 2010 )

PAGE 23

23 Coffee Roasting General Aspects Roasting is a step of the coffee processing where the beans are subjected to heat treatment at high temperatures of up to 2 6 0 280 C. Many physical changes occur in the bean and many chemical reactions take place, producing hundreds of chemical compounds th rough several mechanisms ( Buffo and Cardelli Freire 2004 ; Murkovic 2004 ; Franca and others 2009 ) These changes are critical for the development of the aroma, color and final characteristics of the coffee. Over 8 00 volatile s have been identified in the aroma of roasted coffee and most of these chemicals are induced from the roasting process ( Lee and Shibamoto 2002 ) The chemistry of flavor development during coffee roasting is highly complex and not complete ly understood. Even though roasting appears to be simple in terms of processing conditions, it is quite complex from a chemistry point of view, since hundreds of chemical reactions take place simultaneously. Examples include Maillard and Strecker reactions and degradation of proteins, polysaccharides, trigonelline and chlorogenic acids. Sugars, of which sucrose is the most abundant, will act as aroma precursors, resulting in several substances (furans, aldehydes, carboxylic acids, etc.) that will affect bo th flavor and aroma of the beverage ( Schenker and others 2002 ; Farah and others 2006a ) Roasting conditions will also greatly influence the formation of acrylamide, a low molecular compound that has been classified as probably carcinogenic in hu mans ( Lukac and others 2007 ; Summa and others 2007 ) Acrylamide is formed from the amino acid asparagine and reducing sugars, readily available in green coffee. Roasting

PAGE 24

24 t emperatures h igher than 120C favor the formation of acrylamide while temperatures higher than 200C contribute to its degradation ( Gokmen and Senyuva 2006 ) The Process The coffee roasting process consists essentially of applying dry heat to green beans typically at temperatures from 200 C to 260 C for a certain period of time. Although very simple at first, r oasting coffee is a fairly complex process From a technical point of view, coffee roasting is comprised of several parameters and processes influencing each oth er. Heat is transferred from the roaster to the beans predominantly by conduction, through the roaster walls, and by convection, through the air inside the roaster. Mass is also transferred from the beans as water vapor, gases and volatile compounds produc ed from the many exothermic reactions (Figure 2 2) ( Eggers and Pietsch 2001 ) Heat and Mass Transfer Roasting is a time and temperature dependent process. Heat and mass transfer occur at the same time inside coffee beans Heat flows into the bean during the first period of roasting while water vapor leaves the bean in the oppos it e direction. This type of heat transfe r is very ineffective, but inevitable. The complexity of coffee roasting is especially aggravated due to the con tinuous changes in physical parameters throughout the process. During roasting, the coffee bean loses moisture and swells, internal cavities are produced and changes in heat transfer properties take place with in the bean. From a modeling perspective, this process can be complicated A s temperature increases unevenly in the bean, water molecules start to diffuse and are transferred also uneven ly to the bean surface. Th is creates gradient s of mass and heat transfer

PAGE 25

25 inside the bean that will only end when the roasting is concluded and beans are cooled ( Eggers and Pietsch 2001 ; Bonn lander and others 2005 ) Several attempts to measure heat transfer properties in coffee during roasting have been made ( Sivetz and Derousier 1979 ; Raemy and Lambelet 1982 ; Nagaraju and others 1997 ) Specific heat of green and roasted coffee has been measured by heat flow calorimetry technique at 30 C ( Raemy and Lambelet 1982 ) Green arabica specific heat was measured and reported as 1.85 J g 1 C 1 in beans with 7.5% humidity, slightly higher than robusta specific heat of 1.46 J g 1 C 1 containing 4.5 % humidity In r oasted coffee bean it was also 1.46 J g 1 C 1 at 2.5% humidity. The overall heat transfer coefficients ( ) include the heat transfer from gas to bean and conduction into the bean matrix. Values rep orted using drum roasters were o n average 2.84 W m 2 K 1 while in spouted bed roasters they were 14 W m 2 K 1 on average, with the assumption of constant material A fter recalculating using crude data, a value of 10 W m 2 K 1 was found, still indicating a much faster process in spouted bed where heat convection is more predominant, in contrast with conventional drum roasters that heat up primarily by conduction ( Nagaraju and others 1997 ; Eggers and Pietsch 2001 ) Coffee Roasting Methods There are several different technologies available for coffee roasting using the same basic principles of heat and mass transfer Some important methods are described below. Rotating cylinder Using either a horizontal or vertical cylinder, this roasting method has been described as the most traditional in coffee roasting. Beans are mixed in rotating drums

PAGE 26

26 heated by gas burners that t ransfer heat to the beans through the cylinder walls and hot gases. It can be either batch or continuous operated with roasting times varying from eight to twenty minutes in traditional low temperature long time (LTLT) or from three to six minutes in high temperature short time (HTST) roasting systems ( Eggers and Pietsch 2001 ; Mwithiga and Jindal 2003 ) Fixed drum with rotating paddles This method uses direct heating by convective flow of hot gases with roasting times varying from t hree to six minutes. It differs from the rotating cylinder method by using primarily the convective heating source. Also instead of a rotating drum, paddles are used to mix and homogenize the heat transfer inside the roaster ( Eggers and Pietsch 20 01 ; Bonnlander and others 2005 ) Fluidized bed Fluidization of beans is achieved by using high velocity hot air directed towards the beans in a perforated roasting bed. Better control of the process parameters, including heat transfer uniformity, is achieved since beans are floated and mixed simultaneously. Air velocity control is critical in fluidized bed systems as coffee bean size, weight and d ensity drastically change throughout the roasting. Roasting times vary typically from three minutes to five minutes in fast systems ( Eggers and Pietsch 200 1 ) Some variations coming from the fluidized bed system such as spouted bed and swirling bed have also been described and patented for coffee roasting but research is still in need for these alternative methods in order to improve bean temperature homo geneity ( Nagaraju and others 1997 ; Nagaraju and Bhattacharya 2010 ) Figure 2 3 shows the basic schematics of these roasters.

PAGE 27

27 C hemical Composition of Green and Roasted Coffees Carbohydrates G reen coffee chemical composition is dominated by carbohydrates, mostly polysacchar ides ( hemicelluloses and cellulose) forming the basis of cell wall material in the bean S ucrose is the most abundant soluble sugar present typically averaging in concentrations between 3 and 9 % on a dry basis (Table 2 3 ) Arabicas typically contain up to 50% more sucrose than robusta coffees ( Wasserman and others 2000 ; Bradbury 2001 ; Kleinwachter and Selmar 2010 ) M onosaccharides such as glucose, galactose, arabinose, fruc tose, mannose, mannitol, xylose and ribose can be found in concentrations below 1% in green beans ( Arya and Rao 2007 ) while raffinose and stachyose have been reported only in green robusta coffee ( Feldman and others 1969 ) During roasting, most of the sucrose disappears, generating products including low molecular weight acids, 5 hydroxymethylfurfural and some volatile and non volatile flavor compounds formed as a result of Maillard reactions with free and bound amino acids ( Wasserman and others 2000 ) Due to the extensive degradation during roasting, light and dark roasted beans have shown sucrose concentrations of 0.45% and 0.06% respectively in a study with arabica coffees ( Feldman and others 1969 ) Almost all of the hexoses and pentoses are lost, producing water, carbon dioxide, color, aroma and flavor. Cellulose is more stable to roasting ( Parliment 2000 ) Lipids Green coffee contains be tween 7 and 17% lipid material, such as triglycerides sterols (stigmasterol, sitosterol), fatty acids (linoleic, linolenic, oleic, palmitic, stearic, araquidic, lignoceric and behenic), pent acyclic diterpens (methylcafestol, cafestol, kahweol), diterpenic alcohols, diterpenic and tr ipterpenic esters and ceramide. Coffee

PAGE 28

28 wax is located on the surface of the bean and contains C20 and C26 fatty acids ( Parliment 2000 ; Parras and others 2007 ) Table 2 4 shows the composition of the li pid fraction in green coffee. Upon roasting, no ma jor changes have been detected in coffee lipids. However changes such as autoxidation of the unsaturated acids, primarily C18, and hydrolysis of the triglycerides yielding free fatty acids were detected dur ing storage of green and roasted beans ( Parliment 2000 ) Protein The protein content is relatively high in g reen coffee representing between 8.5 and 13% of the coffee bean dry weight. The free amino acids level is r ather low, in the range of 0.3 0.6% on a dry weight basis, and includes amino acids such as asparagine, glutamic acid, ala nine, aspartic acid phenylalanine the sulfur containing methionine and lysine ( Montavon and others 2003 ; Murkovic and Derler 2006 ) Together with carbohydrates amino acids are important precursors of the coffee aroma and flavor Roasting denatures and insolubilizes much of the proteins, and a marked decrease in amin o acids such as arginine, cysteine, lysine, serine and threonine has been reported. Alanine, glycine, leucine, glutamic acid, and phenylalanine are relatively stable to roasting. Cysteine and m ethionine, potential precursors for the many sulfur compounds f ound in coffee aroma, are drastically reduced. ( Wasserman and others 2000 ) Alkaloids Coffee contains several species of xanthines such as caffeine and alkaloids such as trigonelline. There is a slight increase in the relative caffeine content from green to roasted coffees due mainly to the loss of other components ( Merritt and Proctor 1959 )

PAGE 29

29 Caffeine is normally present in the range of 1 .0 to 1.5% in arabica coffees and twice that in robusta s ( Parliment 2000 ) While many studies h ave shown that caffeine is very stable to roasting, trigonelline is degraded by up to 80% during roasting. It generates nicotinic acid, a precursor of many aroma compounds including pyridine and picolines ( Bee and others 2005 ) The decarboxylation of trigonelline is another source of carbon dioxide genera ted during roasting. Typically a rabica c ontains more trigonelline than r obusta coffees Higher trigonelline concentrations are gen erally associated with better quality coffees (Figure 2 4 ) ( Farah and others 2006a ) Chlorogenic A cids Chlorogenic acids are the most abundant category of organic acid s present in coffee Chlorogenic acid or CGA, is a trivial name used to describe a range of phenolic acids esterified to quinic acid found in plant materials, including coffee. The term can in clude at least five groups of isomers of which caffeoylquinic acids (CQA), dicaffeoylquinic acids (diCQA) and feruloylquinic acids (FQA) are the major ones reported in coffee ( Trugo and Macrae 1984 ) Green coffee beans contain the largest amounts of CGA found in plants, ranging from 5 to 12% dry basis ( Farah and others 2005 ) Nine CGAs have been identified in green coffee (Figure 2 5 ) where 5 caffeoylquinic acid ( 5 CQA) represents the most abundant CGA in the green beans accounting for up to 5% of the weight of dry coffee ( Fujioka and Shibamoto 2008 ) Studies on the degradation of some CGA during coffee roasting have been reported using high performance liquid chromatography (HPLC). During roasting, isomerization and degradation of chlorogenic acids occur and part of the CGA is converted into

PAGE 30

30 c hlorogenic acids l actones (CGL) by the loss of a water molecule from the quinic acid and formation of an intramolecular ester bond ( Trugo and Macrae 1984 ; Farah and others 2005 ) S even CGLs have been identified in roasted coffee where caffeoylquinic acid lactones (CQL) are the most abundant. Longer roasting times resulted in lower amounts of both CGA and CGL, showing that CGL are possibly intermediate compounds in the degradation process of CGA at high temperatures ( Farah and others 2005 ; Farah and others 2006b ) Table 2 5 shows the content of CGA in green and roasted coffee from di fferent origins, including two a rabicas and one r obusta coffee. Robusta coffees typ ically contain higher concentrations of CGA, which together with the higher caffeine content, contribute to produce brewed coffees with greater bitterness. Since extensive degradation occurs to CGA during roasting, possible correlations between the concent rations of CGA and darkness of the roast ( Feldman and others 1969 ; Trugo and Macrae 1984 ) and characterization of commercial species ( Clifford and Jarvis 1988 ; Clifford and others 1989 ; Bicchi and others 1995 ) have been in vestigated The ratio of 5 CQA : caffeine has been reported as a n alternative monitor for the degree o f roast instead of inspecting color in coffee roasting. Since 5 CQA is degraded and caffeine is more stable under roasting temperatures, the ratio has show n a good correlation with the degree of roast (Fig ure 2 6 ) Moreover, coffee roasted to the same color but under different roast conditions have presented different ratios of these compounds It has been concluded that monitoring the ratio 5 CQA/caffeine may give a better correlat ion with the actual flavor and aroma profiles ( Purdon and McCamey 1987 )

PAGE 31

31 Organic Acids After c hlorogenic acids, citric acid typically represents the next highest acid concentration in green beans. Malic and quinic acids are present in similar concentrations varying from 1.6 to 7.3 g kg 1 (Table 2 6 ). Several other organic acids have been reported i n green coffees including phosphoric, tartaric, pyruvic, succinic, and acetic acids among the most predominant ( Balzer 2001 ) During roasting, citric and malic acids undergo thermal degradation Citric yield s mainly citraconic, glutaric, itaconic, mesaconic an d succinic acids, while malic generates fumaric and maleic acids. Acidity in coffee has been observed in the roasting process, indicated by a drop in extract pH (Figure 2 7 ). The acidity is typically generated in the first stage of the roasting process followed by a small decrease as roasting progresses ( Balzer 2001 ) Q uinic acid is already present in green coffee. During roasting, a slight in crease in quinic acid has been reported due to chlorogenic acid breakdown into compounds including chlorogenic acid lactones, quinic acid lactones, and quinic acid However it also undergoes degradation produc ing phenol, catechol, hydroquinone and pyrogallol. Diphenols found in coffee aroma are believed to have q uinic and caffeic acids as precursors. Quinic acid lactones also known as quinides, a non sour compound produced from chlorogenic acids under roasting temperatures are hydrolyzed to quinic acid upon coffee brewing That helps explain the increase d sourness of brewed coffees upon standing ( Sivetz 1963 ; Parliment 2000 )

PAGE 32

32 Physical Changes in Coffee during Roasting Moisture Green coffee beans are typically dried to below 12% moisture, to reduce chemical and microbial degradation that would ultimately cause severe off flavor development ( Parliment 2000 ) During roasting b oth free and chemically bound water decrease significantly to about 2 %, depending on the roast degree and whether water quenching is applied immediately after roasting W ater content decreases more rapidly duri ng the first stages of roasting ( Eggers and Pietsch 2001 ) Surface and Structure of the Bean Coffee beans surface color chang e drastically from pale green to brown during roasting. Coffees submitted to dark roasting levels, tend also to sweat oil, rendering a shiny bean surface. The roasting induced formation of color and flavor compounds is accompanied by a large loss of water, carbon dioxide and organic mass. As a consequence, an increase in internal pressure happens and causes severe changes in bean porosity and volume. Typically coffee beans increase at least 50% of their original volume. The pore structure highly dependent on the roasting conditions, controls the mass transfer phenomena during storage and determines the degassing properties of the beans ( Ortola and others 1998 ; Schenker and others 2000 ) Beans show greater increase in volume under fast roasting conditions also called High Temperature Short Time (HTST) in contrast with Low Temperat ure Long Time (LTLT) (Figure 2 8 ). With those severe changes in bean volume and porosity, density reduction also takes place. Raw bean density is typically in the range of 550 to 700 g L 1 while in roasted beans it is from 300 450 g L 1 with lowest figures found for the fast roasting processes ( Bonnlander and others 2005 )

PAGE 33

33 Color Melanoidins are yellow to dark brown heterogeneous polymeric products that occur in many processed foods. They greatly influence sensory and nutritional quality of foods, mainly through their colo r, flavor and antioxidant properties. It has been reported that phenolic polymerization and Maillard reaction are the major contributing reactions to the formation of colored polymer, melanoidin, in roasted coffee ( Borrelli and others 2002 ; Montavon and others 2003 ) The molecular mass of these melanoidins ha s been estimated by HPLC and ranged from 3,000 to 100 ,000 Daltons depending on the coffee species and on the degree of roast, and increased with longer roasting times ( Homma 2001 ) Some later studies have also detected the presence of high concentration of polymers above 100 kDa in coffee brews ( Borrelli and others 2002 ; Gniechwitz and others 2008 ) The most obvious physical change to occur during coffee roasting is color. In large scale commercial roasting, color is typically the monitoring paramet er to measure the degree of roast. Roasted coffee color varies from a light brown color to almost black in very dark roasts. It is typically measured by grinding the roasted beans to a powder of predetermined particle size. The material is distributed onto a disk surface and the surface color reflectance is measured using instruments such as Agtron, Minolta and Hunterlab. Other roast degree indicators are the sound of coffee, especially those emitted during the crackling stages, dry matter loss, and final t emperature, however the most used and accepted reference for measuring the degree of roast in the industry still rely on color measurement ( Parliment 2000 ) There are quite a few different scales and terminologies used to describe the coffee degree of roast. The most used in the coffee industry is the Agtron scale based

PAGE 34

34 on the color measured using Agtron spectrophotometers that carry a proprietary color and many others are still frequently used in the coffee business (Table 2 7 ) ( Davids 2003 ) Coffee Aroma and Flavor Development D uring Roasting Along with its stimulato ry effect, coffee is appreciated for its pleasant aroma and flavor developed primarily during roasting, a time temperature dependent process in which beans undergo a series of reactions leading to several changes in chemical composition. Numerous investiga tions have been carried out to identify the volatile compounds responsible for the roasted coffee aroma and flavor. Pioneer ing research o n coffee volatiles date s from 1920 30 done by Reichstein and Staudin g er ( Reichstein and Staudinger 1926 ) They identified 29 components including the important classes of compounds of alkyl pyrazines, diketones and furfuryl mercaptan, all of that before the days of gas chromatography/mass spectroscopy. The y also concluded that none of those substances caused coffee aroma, except a highly diluted aqueous solution of 2 furfurylthiol ( Flament 2001 ; Grosch 2001 ) Only 13 This number increased to 60 in the following 50 years and to more than 600 constituents from 1965 to 1975, time when progress in instrumental analysis and mass spectrometry has demons trated that the volatile fraction of green and roasted coffee consisted of a great variety of compounds class ( Tressl 1989 ) Today over 200 and 800 volatile compounds have been reported in green and roasted coffee r espectively (Table 2 8)

PAGE 35

35 Only p art of the native green bean volatiles contributes to the final roasted coffee aroma. Most of the aromatic character of coffee is the result of roasting. Many complex physical and chemical changes occur during this process. In the first stage of roasting, free water is lost; in the second stage, complex chemical dehydration, fragmentation, recombination, and polymerization reactions occur. Several of these changes are associated with the Maillard reaction and lead to the formation of low molecular weight compounds, such as carbon dioxide and free water, and the various flavor and aroma compo unds. Roasted coffee contains a very complex mixture of volatile compounds, with concentrations over a very broad range, which makes difficult both identification and quantification of aroma active compounds. ( Parliment and Stahl 1995 ) Table 2 9 shows various volatiles found in roasted coffee divided into groups of similar odor qualities and the concentration range found in a blend of arabica coffee from Colombia The aroma profiles of arabica and robusta coffees are quite different. The roasty earthy, and phenolic notes are more intense in robustas, and sweetish and green notes are more predominant in arabicas. The volatile content of phenols is typically much higher in robustas. Concentration s of guaiacol, 4 ethylguaiacol and 4 vinylguaiacol were significantly higher in robustas in a study comparing medium roasted robustas from Indonesia and arabicas from Colombia P yrazines were also significantly more concentrated in robustas while furanones in arabicas ( Semmelroch and others 1995 ) Table 2 10 shows these numbers in terms of Odor Activity Value (OAV) a unitless measure of importance of a specific compound to the odor of a sample calculated as the ratio between concentration of the individual compound and its odor threshold concentration

PAGE 36

36 The degree of roast and the roasting conditions greatly influence the aroma profile in coffee beans. R oasting coffee from light to dark increases the sulphurous/roasty, earthy and smoky notes in the aroma profile. Analysis of potent odorants suggested that the stronger sulphurous/roasty and smoky notes of the dark sample might be caused by 2 furfurylthiol and guaiacol. Table 2 1 1 shows the increase in many components in arabica coffee from Colombia roasted to light medium and dark roast s including a 110% increase in 2 furfurylthiol and an almost 3 times increase in guaiacol ( Mayer and others 1999 ) Compared to other roasted foods, sulfur constituents and phenols are formed in high amounts in coffee roasting, contribut ing to desirable coffee flavor ( Tressl 1989 ) Findings from a sensory study on the character impact odorants of a roast e d coffee from Colombia showed that the flavor profile of coffee is mainly caused by 2 furfurylthiol, 4 vinylguaiacol, several alkyl pyrazines, furanones, acetaldehyde, propanal, and some Strecker aldehydes. In contrast to 2 furfurylthiol, the other sulfur compounds only had a limited influence on coffee flavor ( Czerny and others 1999 ) Formation of Aroma Compounds Green coffee contains high amounts of chlo rogenic acids (CGA), including three isomers of caffeoylquinic acid three of feruloylquinic acid and three of di feruloylquinic acid found in roasted beans, including guaiacol and catechol. Both guaiacol and its corresponding derivative 4 vinyl guaiacol possess low thresholds and are contributing flavor com ponents in roasted coffee ( Tressl 1989 ) Model experiments indicated that 4 feruloylquinic acid is the likely precursor for guaiacol, 4 ethylguaiacol and 4

PAGE 37

37 vinylguaiacol. Robustas contain more of this acid than arabi ca, thus the higher level of these phenolics is explained. Sucrose, reducing sugars, free and peptide amino acids are important precursors to aroma compounds in coffee. Model experiments revealed that pentoses were significantly more effective than hexoses as precursors of 2 furfurylthiol. Model roasting using arabinogalactan, which is present in large amounts in green coffee, produced 2 furfurylthiol. It was suggested that arabinose from the side chain of arabinogalactan is the likely precursor for furfura l, a dehydration product of pentoses, and the likely immediate precursor to the formation of 2 furfurylthiol. Regarding the sulfur precursors containing the mercapto functional group, a c ysteine containing peptide was more effective than free cysteine ( Parliment and Stahl 1994 1995 ) Coffee Aroma Changes D urin g Storage During storage, the aroma of coffee is not stable. Moisture, oxygen, temperature and light accelerate the development of a stale aroma. Increased concentration of hexanal during storage has been reported due to autoxidation of linoleic acid ( Marin and others 2008 ) Other compounds are prone to loss during storage Sulfur compounds such as 2 furfurylthiol, methional and methanethiol were significantly reduced in a vial headspace by 23%, 29% and 66% respectively after 30 minutes of storage at room temperature ( Grosch and Mayer 2000 ) Other odorants were also significantly reduced (Table 2 12).

PAGE 38

38 Table 2 1. Coffee production for the 6 largest producing countries (in thousands of 60 kg bags). Crop year COUNTRY Species* 2007 2008 2009 2010 Brazil A/R 36,070 45,992 39,470 48,095 Vietnam A/R 16,467 18,500 18,000 18,000 Indonesia A/R 7,777 9,612 11,380 9,500 Colombia A 12,504 8,664 8,500 9,000 India A/R 4,460 4,371 4,827 5,000 Ethiopia A 4,906 4,350 4,500 7,450 Other producing countries A/R 36,798 36,542 32,758 37,078 WORLD PRODUCTION A/R 119,276 128,377 119,894 134,498 A = arabica coffee, R = robusta coffee (Source: ICO 2010a). Table 2 2. Green coffee price indicator in February 2011 (in cents of US dollars per kg). Coffee NY Germany Daily avg Colombian Mild Arabicas 635.5 660.8 649.4 Other Mild Arabicas 615.4 623.5 620.2 Brazilian Naturals 503.2 532.5 525.8 Robustas 248.6 234.1 236.5 (Source: ICO 2010b).

PAGE 39

39 Table 2 3. Sugar content of green arabica and robusta coffees. Coffee Sucrose Fructose Glucose Mannose Arabinose Total Arabica Colombia 8.2 0.15 <0.01 ND <0.01 8.35 Colombia 8.3 0.07 0.3 ND 0.05 8.72 Salvador 7.3 0.02 <0.01 ND 0.09 7.43 Brazil 6.65 0.15 <0.01 0.02 0.15 6.87 Brazil 6.3 0.15 <0.01 0.1 0.07 6.62 Kenya 8.45 0.02 <0.01 <0.01 0.07 8.55 Kenya 7.05 0.03 <0.01 0.06 0.07 7.21 Tanzania 7.55 0.2 0.45 0.08 0.05 8.33 Ethiopia 6.3 0.4 0.4 ND <0.01 7.1 Ethiopia 6.25 0.25 0.45 ND 0.04 6.99 New Guinea 7.7 0.07 <0.01 ND 0.06 7.84 East India 6.5 0.04 <0.01 ND 0.1 6.64 Robusta Madagascar 3.9 0.25 <0.01 ND 0.12 4.29 Cameroun 3.2 0.3 <0.01 ND 0.09 3.6 Cote d'lvoire 3.4 0.35 0.2 ND 0.09 4.06 Cote d'lvoire 0.9 0.55 0.5 ND 0.15 2.1 Indonesia 1.25 0.25 <0.01 ND 0.05 1.56 Indonesia 3 0.2 0.35 0.06 0.07 3.68 Philippines 4 0.4 0.35 0.02 0.1 4.87 Philippines 4.85 0.35 0.5 ND 0.04 5.74 Expressed as % dry basis (Source: Bradbury 2001). Table 2 4. Typical content of the lipid fraction in green coffee. Constituent Content (%) triglycerides 70 80 free fatty acids 0.5 2.0 diterpene esters 15 19 triterpenes and sterols 1.4 3.2 coffee wax 0.25 (Source: Parliament 2000).

PAGE 40

40 Table 2 5. Chlorogenic acid content in green and roasted C. arabica Cv. Bourbon (1), C. arabica Cv. Longberry (2), and C. canephora Cv. robusta (3). Cultivar roast degree 3 CQA 4 CQA 5 CQA FQA di CQA Total CGA 1 green 483 544 3126 279 770 5202 Very light 996 1278 2809 315 507 5904 light 816 1000 1996 225 336 4373 light medium 459 590 1104 165 137 2455 dark medium 199 282 448 114 43 1085 dark 123 156 244 52 22 596 very dark 80 71 160 34 2 347 2 green 478 474 3606 292 840 5690 very light 951 1234 3102 294 647 6228 light 805 933 1724 206 286 3954 light medium 344 495 816 152 131 1937 dark medium 181 265 376 80 38 941 dark 110 156 219 41 15 540 very dark 53 51 105 26 0 235 3 green 925 602 4243 471 1336 7577 Very light 1257 1481 3802 509 951 8000 light 1087 1343 2521 409 581 5940 light medium 623 791 1347 335 235 3332 dark medium 334 308 448 184 60 1335 dark 219 168 226 102 19 733 very dark 108 100 111 56 8 383 Results expressed in mg/100g of coffee. 3 CQA = 3 caffeoylquinic acid; 4 CQA = 4 caffeoylquinic acid; 5 CQA = 5 caffeoylquinic acid; FQA = feruloylquinic acids comprising of 3 FQA, 4 FQA and 5 FQA; diCQA = dicaffeoylquinic acids comprising of 3,5 diCQA, 3,4 diCQA, and 4,5 diCQA. (Source: Farah a nd others 2005). Table 2 6. Acid contents in different green arabica and robusta coffee samples analyzed by HPLC/UV. acid Arabica Robusta Citric 5.0 14.9 3.3 10.1 Malic 2.6 6.7 1.8 7.3 Quinic 3.3 6.1 1.6 8.6 Succinic tr 1.5 0.5 3.5 Formic tr 1.4 tr 3.9 Acetic tr tr 2.0 Results for arabica and robusta coffees contain a range of 8 and 7 samples respectively, expressed in g/kg dry basis. tr = trace amounts. (Source: Balzer 2001).

PAGE 41

41 Table 2 7. Guide used in color roasting with information about the end of roast average temperatures, Agtron color and some of the common descriptors used for coffee roasts. Roast Color Final Bean Temp erature (C) Agtron Gourmet Scale Agtron/ SCAA classification Common Names Very light brown Around 195 95 90 Tile #95 Cinnamon Light brown Below 205 90 80 Tile #85 Cinnamon, New England Moderately light brown Around 205 80 70 Tile #75 Light, New England Light medium brown 205 215 70 60 Tile #65 Light Medium American, Regular Brown Medium brown 215 225 60 50 Tile #55 Medium, Medium high American, Regular City Medium dark brown 225 230 50 45 Tile #45 Viennese, Full City, Light French Espresso, Light Espresso, Continental, After dinner Moderately dark brown 230 235 45 40 Espresso, French, European High Dark brown 235 240 40 35 Tile #35 French Espresso, Italian, Dark Turkish Very dark brown 240 245 35 30 Italian, Dark French, Neapolitan, Spanish, Heavy Very dark (nearly black) 245 250 30 25 Tile #25 Dark French, Neapolitan, Spanish (Source: Davids 2003)

PAGE 42

42 Table 2 8. Number of volatile compounds reported in green and roasted coffee. Class of compound Green coffee Roasted coffee Hydrocarbons 41 80 Alcohols 24 24 Aldehydes 32 37 Ketoses 21 85 Carboxylic acids 3 28 Esters 26 33 Nitrogen compounds 38 224 Pyrazines 86 Pyrroles 66 Pyridines 20 Other N compounds 52 Sulphur compounds 7 100 Furanes 17 126 Phenols 10 49 Oxazoles 35 Others 9 20 Total 228 841 (Source: Parliment and Stahl 1995; Grosch 2001)

PAGE 43

43 Table 2 9. Concentrations in medium roasted arabica coffee from Colombia divided into groups of volatile compounds with similar odor qualities. Group / O dorant Concentration range (mg/kg) Sweetish/caramel Methylpropanal 24.0 32.3 2 Methylbutanal 20.7 26.0 3 Methylbutanal 17.0 18.6 2,3 Butanedione 48.4 50.8 2,3 Pentanedione 34.0 39.6 4 Hydroxy 2,5 dimethyl 3(2H) furanone 112 140 Ethyl 4 hydroxy 2 methyl 3(2H) furanone 16.0 17.3 Vanillin 3.4 4.8 Earthy 2 Ethyl 3,5 dimethylpyrazine 0.249 0.400 2 Ethenyl 3,5 dimethylpyrazine 0.052 0.053 2,3 Diethyl 5 methylpyrazine 0.073 0.100 2 Ethenyl 3 ethyl 5 methylpyrazine 0.015 0.018 3 lsobutyl 2 methoxypyrazine 0.059 0.120 Sulphurous/roasty 2 Furfurylthiol 1 .68 1 .70 2 Methyl 3 furanthiol 0.060 0.068 Methional 0.228 0.250 3 Mercapto 3 methylbutyi l formate 0.077 0.130 3 Methyl 2 buten 1 thiol 0.0082 0.01 3 Methanethiol 4.4 4.7 Dimethyl trisulphide 0.028 Smoky/phenolic Guaiacol 2.4 4.2 4 Ethylguaiacol 1 .42 1 .8 4 Vinylguaiacol 45 65 Fruity Acetaldehyde 120 139 Propanal 17.4 (E) Damascenone 0.195 0.260 Spicy 3 Hydroxy 4,5 dimethyl 2(5H) furanone 1.36 1.90 4 Ethyl 3 hydroxy 5 methy l 2(5H) furanone 0.104 0.160 (Source: Grosch 2001).

PAGE 44

44 Table 2 10. Odor thresholds and Odor Activity Values (OAV) of some important odorants from roasted samples of arabica and robusta coffees. Odorant Odor threshold OAV Arabica Robusta 2 Furfurylthiol (1) 0.01 1.1 x 10 5 1.7 x 10 5 2 Ethyl 3,5 dimethylpyrazine (2) 2 165 470 2,3 Diethyl 5 methylpyrazine (3) 1 95 310 (E) Damascenone (4) 0.00075 2.6 x 10 5 2.7 x 10 5 Methional (5) 0.2 1.2 x 10 3 0.5 x 10 3 3 Mercapto 3 methylbutyl formate (6) 0.0035 3.7 x 10 4 3.3 x 10 4 Guaiacol (7) 2.5 1.7 x 10 3 1.1 x 10 4 4 Vinylguaiacol (8) 20 3.2 x 10 3 8.9 x 10 3 4 Ethylguaiacol (9) 50 32 362 Vanillin (10) 25 192 644 4 Hydroxy 2,5 dimethyl 3(2H) furanone (11) 10 1.1 x 10 4 5.7 x 10 3 3 Hydroxy 4,5 dimethyl 2(5H) furanone (12) 20 74 32 5 Ethyl 3 hydroxy 4 methyl 2(5H) furanone (13) 7.5 21 11 5 Ethyl 4 hydroxy 2 methyl 3(2H) furanone (14) 1.15 1.5 x 10 4 1.2 x 10 4 (Source: Semmelroch and others 1995). Table 2 11. Odor Activity Values (OAV) of some important odorants from a Colombian arabica coffee roasted to light (l), medium (m) and dark (d) degree of roast. Odorant OAV Colombia (l) Colombia (m) Colombia (d) 2 Furfurylthiol (1) 1.31 x 10 5 1.68 x 10 5 2.74 x 10 5 2 Ethyl 3,5 dimethylpyrazine (2) 132 124.5 159 2,3 Diethyl 5 methylpyrazine (3) 95 73 113 (E) Damascenone (4) 2.93 x 10 5 2.96 x 10 5 3.16 x 10 5 Methional (5) 1.25 x 10 3 1.14 x 10 3 1.52 x 10 3 3 Mercapto 3 methylbutyl formate (6) 1.06 x 10 4 2.2 x 10 4 2.23 x 10 4 Guaiacol (7) 812 1216 2212 4 Vinylguaiacol (8) 2.78 x 10 3 2.76 x 10 3 2.85 x 10 3 4 Ethylguaiacol (9) 17.4 28.4 49.4 Vanillin (10) 116 136.4 146 (Source: Mayer and others 1999).

PAGE 45

45 Table 2 12. Loss of potent odorants from ground roasted Colombian coffee at room temperature. Group / Odorant Loss (%) Sweetish/caramel group Methylpropanal 25 2 Methylbutanal 32 3 Methylbutanal 27 2,3 Butanedione 19 2,3 Pentanedione 25 4 Hydroxy 2,5 dimethyl 3(2H) furanone 1.4 2 Ethyl 4 hydroxy 5 methyl 3(2H) furanone 1.3 Vanillin 20 Earthy group 2 Ethyl 3,5 dimethylpyrazine 12 2 Ethenyl 3,5 dimethylpyrazine 6.6 2,3 Diethyl 5 methylpyrazine 13 2 Ethenyl 3 ethyl 5 methylpyrazine 13 3 lsobutyl 2 methoxypyrazine 21 Sulphurous/roasty group 2 Furfurylthiol 23 Methional 29 Methanethiol 66 Smoky/phenolic group Guaiacol 18 4 Ethylguaiacol 8.4 4 Vinylguaiacol 4.9 Fruity group Acetaldehyde 45 (E) Damascenone 12 Spicy group 3 Hydroxy 4,5 dimethyl 2(5H) furanone 1.1 (Source: Grosch and Mayer 2000).

PAGE 46

46 Figure 2 1. Coffee cherry longitudinal section showing the cross section of the beans, silverskin, pulp, parchment and skin. (Source: Arya and Rao 2007). Figure 2 2. Heat and mass transfer aspects during coffee bean roasting. ( Source: Eggers and Peitsch 2001).

PAGE 47

47 Figure 2 3. Basic schematics in different roasting technologies. (Source: Bonnlander and others 2005; Nagara ju and others 1997).

PAGE 48

48 Figure 2 4. Trigonelline content in green and roasted coffee beans classified by cup quality (cup quality increases from right to left). Soft, Hard, Rioysh, Rio and Rio Zona are grades given to arabica coffees. (Source: Farah and oth ers 2006b).

PAGE 49

49 Figure 2 5 Structure of chlorogenic acids (CGA) identified in coffee ( Source: Fujioka and Shibamoto 2008).

PAGE 50

50 Figure 2 6. 5 CQA/caffeine ratio versus L* values in coffee roasted to different roast degrees. (Source: Purdon and MacCamey 1987). Figure 2 7. Development of titratable acidity as a function of organic roast loss for Colombian arabica ( ) and Indonesian robusta ( ). (Source: Balzer 2001). 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 16 18 20 22 24 26 28 5 CQA/caffeine ratio Color (L* values) 0 20 40 60 80 100 120 0 1 2 3 4 5 6 7 8 Titratable acidity (end pH 6.0) (mmol/kg) Organic roast loss (%)

PAGE 51

51 Figure 2 8. Relative bean volume increase as a function of roast loss under HTST ( ) and LTLT ( ) roasting conditions. (Source: Schenker and others 2000) 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 0 2 4 6 8 10 12 14 16 18 20 Relative bean volume ( ) Roast loss (%)

PAGE 52

52 CHAPTER 3 COFFEE COLOR: KINETICS OF COLOR FORMATION DURING THE ROASTING PROCESS Commercial coffee roasting is typically a non isothermal process where temperature continually increases with time (Figure 3 1) During th is unit operation chemical reactions, such as oxidation, reduction, hydrolysis, polymerization, decarboxylation etc. and major alterations in physical properties including color, volume, weight, pH, volatile co mponents, CO 2 production and moisture loss take place. These complex reactions and physical alterations depend on the time and temperature during roasting ( Houessou and others 2007 ) Color is of extreme importance during coffee roasting. Typically, the degree of roas t of coffee is measured using color of ground beans, although other direct and indirect methods have been proposed ( Shwartzberg 2002 ; Baggenstoss and others 2008a ) C olor of coffee has been reported in scientific papers as lightness values from the L*, a*, and b* color space ( Purdon and McCamey 1987 ; Schenker 2000 ; Baggenstoss and others 2008a ; Franca and others 2009 ; Baggenstoss and others 2010 ) in L*, a*, and b* values ( Schenker and others 2000 ; Gokmen and Senyuva 2006 ) in hue angle ( Pittia and others 2001 ) and in brightness ( Hernandez and others 2008a ) During roasting, the pale green color of raw beans undergo es change s to an intermediate light brown yellow followed by a gradual darkening to the various degrees of roasts typically found in the final product. From a technology perspective, not much information is provided by the majority of commercial roasters, except the temperature of the beans surface, and in some cases the time temperature history of the process in computerized systems. Roast masters rely typically on this information together with

PAGE 53

53 bean color to decide when the roast is However the latest parameter is only obtain ed o ff line through visual inspection of the bean external color and in some instances by assessing the color of ground beans Due to the nature of the roasting process (agitation, high temperature), t he nonexistence of proper instrumentation to measure more parameters on line other than temperature is the limiting factor for a full automation in coffee roasters ( Hernandez and others 2008a ) Kinetics of color changes due to heat processing in foods have been studied by many authors ( Broyart and others 1998 ; Ibarz and others 1999 ; Lau and others 2000 ; Krokida and others 2001 ) and typically, L* value has been taken as a measurement control for browning ( Lozano and Ibarz 1997 ) However there are very few studies on color changes of the roasting processes ( Ozdemir and Devres 2000 ; Demir and others 2002 ; Kahyaoglu and Kaya 2006 ) Brightn ess c hanges during coffee roasting ha ve been recently s tudied u sing an on line color assessment system inside a lab roaster ( Hernandez and others 2008a ) By determining how the grey scale of beans change d with temperature and time they prop osed four consecutive stages of color changes during roasting: (1) Bean color remained constant below temperatures of 100 C. (2) Bean temperature above 100 C, color slightly darker. (3) Bean temperature above 160 C, color markedly lighter. (4) Gradual darkening to reach f inal product color. Kinetic study on coffee color is quite complex. From an engineering point of view, roasting consists of a combined heat and mass transfer superposed by endothermic and exothermic reactions ( Bonnlander and others 2005 ) The beans change from light

PAGE 54

54 green to dark brown color as a result of the combined temperature and time treatment. Since temperature is not constant during roastin g, and many bean factors, such as moisture content, volume, density texture and pressure interfere with bean thermal properties and ultimately heat transfer mechanisms ( Shwartzberg 2002 ) the kinetic study becomes more challenging. The use of ground green coffee under isothermal conditions may be a first step to determine the color kinetic parameters and t o model the color changes under any thermal process conditions Our objective in this study was to determine the se kinetic parameters using ground green coffee under four isothermal conditions. Materials and Methods Sample Preparation Dry processed g reen arabica coffee ( Coffea arabica L.) variety Bourbon from Ipanema Coffees (Alfenas M.G., Brazil) harvested during the 2008 2009 season were pre sorted in bean sizes between 6.35 mm and 6.75 mm and packaged in 60 kg burlap bags. Upon receiving in Gainesville, FL, beans were transferred to 46 cm length x 76 cm height x 0.1 mm thickness polyethylene bags filled with nitrogen, and frozen at 18C. The moisture content of the beans was 9.98% d.b. To obtain ground green coffee, beans were frozen to 80C, to make t hem brittle, and were ground in a burr coffee grinder (Krups GVX2, Medford, MA U.S.A. ) to pass a 425 m sieve (Fisher Scientific, Pittsburgh, Pennsylvania, U.S.A.). Samples of one gram of the prepared ground green coffee were evenly distributed on a 57mm diameter aluminum weigh dish (Fisher Scientific, Hampton, NH) and roasted at either a constant 170C, 180C, 190C or 200C for several different times in a lab oven (Precision, Winchester, VA U.S.A. ). Temperature was controlled manually using a variable

PAGE 55

55 transformer (Staco Energy, Dayton, OH U.S.A. ). A thermocouple placed 1 cm above the samples and connected to a data acquisition system DAQ 56 (Omega, Stamford, CT U.S.A. ) and RealTime software version 1.4.0 (Engineering & Cybersolutions Inc., Gainesville FL U.S.A. ) were used to monitor and record temperature. Treatment times were selected based on preliminary data using the color changes of samples. Oven was preheated to the desired set point temperature. Samples were then loaded into the oven for the several treatment times presented in Table 3 1 Upon removing the samples from the oven, they were placed on a cold metal surface at room temperature for quick cooling. Samples were then stored in a desiccator with silica gel before being analyzed. Color Analysis The color of four different spots of the ground coffee samples was assessed using machine vision. This system consisted of a light box ( Luzuriaga and others 1997 ) a Nikon D200 digital color camera with a Nikon DX 18 200mm VR II Lens (Nikon Corp., Tokyo, Japan) connected to a computer through a USB cable and Lens Eye Software (Engineering & Cybersolutions Inc., Gainesville, Florida, U.S.A.) developed in our lab, using Visual Basic for Windows (Microsoft, Redmond, Washington, U.S.A.). The light box used 2 fluorescent light bulbs (Lumichrome F15W1XX, color temperature = 6500 K, color retention index=98, Lumiram, Larchmont, New York, U.S.A.) emulating the D65 illumination (natural daylight at noon). Diffuse light inside the box was obtained by using a Polycast acrylic nr 2447 plastic sheet (Faulkner Plastics, Gainesville, Florida, U.S.A.) between the fluorescent bulbs and the sample space. The Nikon camera settings listed in Table 3 2 were obtained in preliminary studies by minimizing the color difference of four Labsphere (North Sutton, New Hamshire, U.S.A.) color standards (red, blue, green

PAGE 56

56 and yellow) Color analysis was performed using Lens Eye software. The roasted c offees, filled into glass Petri dishes (87 mm in diameter and 12 mm in height), were placed individually at the bottom of the light box and a picture was captured from the digital camera set on a tripod facing the sample at the bottom of the light box. The captured images (1000 x 669 pixels) taken with the MV system were calibrated with a Labsphere standard red plate (L* = 48.62, a* = 49.04, b* = 25.72). The images were 24 bit color, meaning that in the Red, Green, Blue (RGB) color space, each color axis wa s represented by 8 bits or 28 = 256 different values. Using the LensEye software, first the RGB values of every pixel of a sample image were read, then this color information was converted to the L*, a*, and b* values, and averaged for each sample image. T his in lightness ( L) in redness ( a* ) and in yellowness ( b* ) w ere calculated from the L*, a*, b* values using e quation s 3 1 3 2, 3 3 and 3 4, respectively ( 3 1) ( 3 2 ) ( 3 3 ) ( 3 4 ) Where x and y refer to the sample s being tested. Kinetic Considerations The complexity of color formation during roasting implies a wide range of reactions caused by the thermal process Thus, it is difficult to establish one single reaction mechanism to obtain a kinetic model describing the process.

PAGE 57

57 L* values Using a graphica l method, linear regression analysis was performed on the plot of L* values versus time for the z ero, half, first and second order reactions. The best fitted line was decided by observing the highest coefficient of determination (r 2 ) ( Hill and Griegerblock 1980 ) T he rate of change o f L* values was expressed by e quation 3 5 (3 5 ) w here L is L* value, t is time, k the rate constant and n is the kinetic order of the reaction (n = 0 is zero order, n = 0.5 is half order, n = 1 is first order, and n = 2 is seco nd order) The rate constant is temperature dependent and often follows Arrhenius relationship ( Labuza and Riboh 1982 ; Taoukis and others 1997 ) expressed as: (3 6 ) where k 0 is the reaction rate constant, E a is the activation energy (J mol 1 ) R is the universal gas constant (8.3144 J mol 1 K 1 ) and T is the absolute temperature in Kelvin. a* and b* values Irreversible reaction in s eries was considered for a* and b* values. This can be written as: The first order linear differential equation applied to this reaction type was modified by adding the initial value C 1 to the expression (3 8)

PAGE 58

58 Where y is the value to be predicted (a* or b*), k 1 and k 2 are the reaction rate constants and C 1 and C 2 the parameters found by curve fitting Optimiz ed values of the four equation parameters (C 1 C 2 k 1 and k 2 ) at the four isothermal temperatures w ere found using TableCurve 2D (Systat Software, Inc., San Jose, Calif., U.S.A.) These parameters were plotted versus temperature and were graphically fitted into linear or exponential equations. Statistical Analysis All statistical analyses were performed in SAS version 9.1 (SAS Institute Inc., Cary, N.C., U.S.A.). Kinetics parameters were analyzed using analysis of variance (ANOVA). Differences between values were determined by mean comparison using Tukey test ( = 0.05). Results Thermal T reatments Figure 3 2 depicts an image containing some of the samples treated at 180 C for different times. Time temperature p l ots of the isothermal treatments at temperatures of 170, 180, 190 and 200 C used in this study are presented in Figure 3 3 Color of the samples was immediately assessed using a MV system L* values Plots of zero, half, first and second order reactions for L* values are presented in Figure 3 4 The results from graphical determination of reaction order are summarized in Table 3 3. Although second order reaction obtained higher r 2 first order was also considered for L* values. The reaction rate constant was determined from the slope of the curves (Table 3 4) and finally plots of Ln k versus T 1 were obtained to determine

PAGE 59

59 temperature dependence of the rate constant. First and second order kinetics followed Arrhenius relationship with r 2 of 0.9976 and 0.9985 r espectively (Figure 3 5). Activation e nergy (E a ) was then calculated from the slope of the Arrhenius plots. E a for L* values for first and second order reactions were 128.7 KJ mol 1 K 1 and 145 .9 KJ mol 1 K 1 respectively and were significantly differen t ( = 0.05) Activation energy for non enzymatic browning in foods are generally between 37.7 and 167.5 kJmol 1 ( Demir and others 2002 ) The predicting first or der reaction model was compared to the experimental data in isothermal conditions at 170, 180, 190 and 200 C (Figure 3 6). The prediction error was calculated by determining the sum of squares (SS) of predicted versus experimental. SS were 64.4, 25.9, 16.8 and 16.3 at 170 180, 190 and 200 C respectively. The same model was tested for whole coffee bean roasting under non isothermal conditions (Figure 3 7). Better p rediction s of L* values w ere accomplished with this model at the later stages of roasting H ow ever relatively high L* w ere found between predicted and experimental data Since many other variables play a significant role during roasting of whole beans, such as temperature distribution within the coffee bean ( Hernandez and others 2007 ) and /or exothermic reactions that occur typically above 170 C and increase with temperature ( Raemy and Lambelet 1982 ) the actual rate may differ from the model system used (ground green coffee). T he second order reaction model was then tested for whole coffee bean roasting under non isothermal conditions (Figure 3 8). The fitting observed was visually better in the second order reaction model and was finally compared to first order reaction model by calculating SS which for the f irst and second order reaction models were 59.5 and 30.8 respectively. By eliminating part of

PAGE 60

60 the initial stages of roasting, where o nly intermediate products were present, and using the 5 later roasted samples, SS were 27.3 and 10.4 for first and seco nd order reaction models respectively. On average, L* values calculated deviated from experimental values by 2.56 and 4.96 for first and second order models respectively. a* and b* values Plots of a* and b* values of the samples treated at the four studied temperatures versus time are presented in Figure 3 9. The initial variation in a* and b* values showed sharp increases in redness and yellowness of samples, followed by slower decrease in these two color attributes. a* remained relatively higher than their initial values, but b* decreased to below green coffee levels. Th e behavior observed for a* and b* were considered similar to irr eversible reactions in series, and for this reason, reaction of this type was considered for both a* and b* values. The model parameters obtained from curve fitting optimization are presented on Table 3 5. These values were fitted to linear or exponential equations from the ir plot versus temperature. The predicting model was then compared to the experimental data in isothermal conditions at 170, 180, 190 and 200 C (Figure s 3 10 and 3 11 ). The prediction error at each isothermal temperature studied, express ed in sum of squares (SS) and the average a* and b* between predicted and experimental are presented in Table 3 6 Very low SS were observed for the fitting of a* and slightly higher for b* values. On average a* and b* deviated from experimental values by 0.68 and 1.88 respectively. The more complex color change mechanisms presented difficulties for modeling a* and b* under non isothermal conditions. Among the kinetics parameters only k 2 observed Arrhenius relationship. Although other empirical models can be used to determine the temperature dependence of these factors ( van Boekel 2009 ) the number

PAGE 61

61 of temperatures studied, especially higher than 200 C, hindered the possibility to obtain a prediction of the change in color E, by combining the three color components kinetics into one model. Summary Change in L* values in ground green coffee roasting, seems to follow either 1 st or 2 nd order kinetics. However 2 nd order reaction better predicted L* values in whole bean roasting, due to factors discussed in this chapter The energy of activation for 1 st and 2 nd order reaction w ere 128.7 KJ mol 1 and 145.8 KJ mol 1 respectively. The rate constant s k 0 were 5.5 x10 1 1 sec 1 and 2.1x10 12 L 1 sec 1 The temperature dependent rate k followed Arrhenius relationship. Although L* is just one component of color, this parameter is probably the most important in roasting. Except for a very small increase for values in whole bean roasting, L* decrease d with temperature and time, this is why many authors anchor the color degree of roast with this color attribute. Changes in a* and b* values, in gro und green coffee roasting, seem to follow a two stage mechanism that can be analyzed as irreversible reactions in series. The model showed good agreement with experimental data under isothermal conditions. Because some kinetic parameters presented great va riability, further studies to obtain these parameters at higher temperatures are needed However much shorter sampling times will arise from those treatment conditions and other techniques should be implemented to make this approach possible.

PAGE 62

62 Table 3 1. List of treatment temperatures and times. Roasting temperatures 170C 180C 190C 200C Roasting times (sec) 0 0 0 0 300 240 180 120 600 480 270 150 900 720 360 180 1200 960 450 210 1500 1200 540 240 2100 1440 630 300 2700 1680 720 360 3600 1920 810 420 5400 900 480 540 600 Table 3 2 Settings used in the Nikon D200 camera. Setting Exposure Mode f/11 Shutter Speed 1/3 sec Aperture 0 EV* Exposure Compensation 0 EV* Sensitivity (ISO) 100 Color temperature Direct sunlight Hue adjustment 0 *Exposure Value Table 3 3. Coefficients of determination (r 2 ) for the plots of zero, half, first and second order kinetics for L* values. Temperature 0 order 1/2 order 1st order 2nd order ( C) L vs. time (L/L 0 ) 0.5 vs. time ln(L/L 0 ) vs. time 1/L vs. time 170 0.76 0.84 0.9 0 0.97 180 0.85 0.92 0.97 0.97 190 0.89 0.95 0.987 0.999 200 0.84 0.92 0.97 0.981 Table 3 4. Rate constant values calculated for first and second order reactions of L* values. T (C) 1 st order k (sec 1 ) 2 nd o r der k (L* 1 sec 1 ) 170 3 73 x 10 4 1 x 10 5 1.39 x 10 5 7.4 x 10 7 180 7 9 1x10 4 2.5x 10 5 3.06 x 10 5 1.9x10 6 190 1 81 x10 3 9.6 x 10 6 7.27 x 10 5 1.4x10 8 200 3 31 x10 3 2.5 x 10 5 1.70 x 10 4 1.6x10 6

PAGE 63

63 Table 3 5. Model parameters found by curve fitting for a* and b* values. Color T (C) C 1 C 2 k 1 k 2 a* 170 1.41 20.40 13.2 3 0.55 8 180 1.09 24.5 8 18.4 8 1.88 190 1.3 9 51.5 2 10.5 8 10.8 7 200 1.41 63.27 16.84 27.79 b* 170 11.97 154.0 6 3.24 16.93 180 9.31 58. 4 12. 4 12.39 190 9.29 58.07 21.95 21.9 3 200 7.2 9 72.0 8 40.96 41.03 Table 3 6. Prediction error calculated in SS, and average difference ( ) for a* and b* values at 170, 180, 190 and 200 C Roasting temp (C) Nr. of o bs ervations SS a* SS b* 170 12 2.4 0.2 24.0 2.0 180 9 6.3 0.7 24.3 2.7 190 10 10.3 1.0 15.6 1.6 200 10 9.0 0.9 13.0 1.3

PAGE 64

64 Figure 3 1. Typical bean time temperature profile in drum coffee roasters. Figure 3 2. Example of image taken from the heat treated ground coffee samples for color analysis. Ground coffee samples (in duplicates) were treated at 180 C for 0, 240, 480, 720, 960, 1200, 1440, 1680, and 1920 seconds. 0.0 50.0 100.0 150.0 200.0 250.0 0 5 10 15 20 Roasting T ( C) Roasting time (min) coffee bean loading

PAGE 65

65 Figure 3 3. T ime t emperature data recorded for samples treated at 170C 180 C, 190 C and 200 C for different sampling times. 100 110 120 130 140 150 160 170 180 190 200 0 2000 4000 6000 Temperature ( C) time (sec) 170 C 300 sec 600 sec 900 sec 1200 sec 1500 sec 2100 sec 2700 sec 3600 sec 5400 sec 100 110 120 130 140 150 160 170 180 190 200 0 500 1000 1500 2000 2500 Temperature ( C) time (sec) 180 C 240 sec 480 sec 720 sec 960 sec 1200 sec 1440 sec 1680 sec 1920 sec 100 110 120 130 140 150 160 170 180 190 200 0 200 400 600 800 1000 Temperature ( C) time (sec) 190 C 180 sec 270 sec 360 sec 450 sec 540 sec 630 sec 720 sec 810 sec 900 sec 100 110 120 130 140 150 160 170 180 190 200 0 200 400 600 800 Temperature ( C) time (sec) 200 C 120 sec 150 sec 180 sec 210 sec 240 sec 300 sec 360 sec 420 sec 480 sec 540 sec 600 sec

PAGE 66

66 Figure 3 4. Plots for zero, half, first and second order of L* values versus time for samples treated at isothermal cond itions at 170 ( ), 180 ( ), 190 ( O ) and 200 C ( + ). 0 10 20 30 40 50 60 70 0 1000 2000 3000 4000 L* time (sec) Zero order 0 0.2 0.4 0.6 0.8 1 1.2 0 1000 2000 3000 4000 (L*/L 0 *) 0.5 time (sec) Half order -2.5 -2 -1.5 -1 -0.5 0 0 1000 2000 3000 4000 ln (L*/L 0 *) time (sec) First order 0 0.02 0.04 0.06 0.08 0.1 0.12 0 1000 2000 3000 4000 L 1 time (sec) Second order

PAGE 67

67 Figure 3 5. Arrhenius plot of first ( ) and second ( ) order reactions for L* values showing the obtained r 2 R = 0.9976 R = 0.9985 -12 -11 -10 -9 -8 -7 -6 -5 -4 0.0021 0.00212 0.00214 0.00216 0.00218 0.0022 0.00222 0.00224 0.00226 0.00228 ln k 1/T (K 1)

PAGE 68

68 Figure 3 6. Plots of predicted ( and solid lines) and experimental ( and dashed lines) L* values versus time at isothermal conditions of 170 180, 190 and 200C, for ground green coffee considering first order reactions. 0 10 20 30 40 50 60 70 0 1000 2000 3000 4000 5000 6000 L* time (sec) 170 C 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 L* time (sec) 180 C 0 10 20 30 40 50 60 70 0 200 400 600 800 1000 L* time (sec) 190 C 0 10 20 30 40 50 60 70 0 200 400 600 800 L* time (sec) 200 C

PAGE 69

69 Figure 3 7. Plot of predicted (solid line) and experimental ( ) L* values versus time at non isothermal conditions, for whole beans roasted in Ambex drum roaster, considering first order reactions. Figure 3 8. Plot of predicted (solid line) and experimental ( ) L* values versus time at non isothermal conditions, for whole beans roasted in Ambex drum roaster, considering second order reactions. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 0 100 200 300 400 500 600 700 800 900 L* Roasting time (sec) 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 0 100 200 300 400 500 600 700 800 900 L* Roasting time (sec)

PAGE 70

70 Figure 3 9. Plots of a* and b* values of the samples treated at isothermal conditions at 170 ( ), 180 ( ), 190 ( O ) and 200 C ( + ) versus time. -5.00 0.00 5.00 10.00 15.00 20.00 0 1000 2000 3000 4000 a* time (sec) 0 5 10 15 20 25 30 35 0 500 1000 1500 2000 2500 3000 3500 4000 b* time (sec)

PAGE 71

71 Figure 3 10. Plots of predicted ( and solid lines) and experimental ( and dashed lines) a* values versus time at isothermal conditions of 170 180, 190 and 200C, for ground green coffee. -5 0 5 10 15 20 0 1000 2000 3000 4000 5000 6000 a* time (sec) 170 C -5.0 0.0 5.0 10.0 15.0 20.0 0 500 1000 1500 2000 2500 a* time (sec) 180 C -5.0 0.0 5.0 10.0 15.0 20.0 0 200 400 600 800 1000 a* time (sec) 190 C -5 0 5 10 15 20 0 200 400 600 800 a* time (sec) 200 C

PAGE 72

72 Figure 3 11. Plots of predicted ( and soli d lines) and experimental ( and dashed lines) b* values versus time at isothermal conditions of 170 180, 190 and 200C, for ground green coffee. 0 5 10 15 20 25 30 35 40 0 1000 2000 3000 4000 5000 6000 b* time (sec) 170 C 0 5 10 15 20 25 30 35 0 500 1000 1500 2000 2500 b* time (sec) 180 C 0 5 10 15 20 25 30 35 0 200 400 600 800 1000 b* time (sec) 190 C 0 5 10 15 20 25 30 35 40 0 200 400 600 800 b* time (sec) 200 C

PAGE 73

73 CHAPTER 4 COMPARISON OF MINOLTA, HUNTERLAB AND MACHINE VISION IN ASSESSING THE COLOR OF ROASTED COFFEE AND THE AGTRON SCAA ROAST COLOR CLASSIFICATION S YSTEM Color is one of the most important attributes of roasted coffee. During roasting, color is used as the indicator of coffee flavor and aroma development ( Little and Mackinney 1956 ; Bonnlander and others 2005 ; Hernandez and others 2008b a ) Before 1995, the coffee industry had no established standards or references for classifying the degree of coffee roast. Unt il then, expert coffee tasters had established individual judgment standards based on color for the roasting of green coffee ( Little and Mackinney 1956 ; Little and others 1958 1959 ; Lockhart 1960 ) In 1995 the degree of roast classification was established by a technical standards committee formed by the Specialty Coffee Association of America (SCAA, Long Beach, California, U.S.A.) in partnership with Agtron Inc. (Reno, Nevada, U.S.A.), a company that designs and manufactures spectrophotometers primarily for the food and beverage industries. In designing the classification, the technical standards committee recognized that the terminol ogy to describe roast color was very subjective. A new vocabulary was developed referring to each of the roasts by number. Thus, a standardized system for describing roast color was established. A consensus of opinion among experienced roasters was used to establish the anchor points for the extremes of classification (the darkest roast color versus the lightest roast color used commercially). Participating SCAA member roasters submitted three samples from three separate roasts that each member produced as their darkest roast. By eliminating the highest and lowest 10% of the samples, the average darkest roast value was set as 30 on the scale, which established the lower limit. In a separate blind cupping conducted by SCAA, samples

PAGE 74

74 were roasted at the indexes of 100, 95, 90, 85, and 80. Seven of the 10 cuppers selected the 90 classification as the lightest roast, establishing the scale's upper limit. From the information obtained from the sensory consensus, the Agtron SCAA Roast Color Classification System (Ag tron RCCS) was established with eight color disks, numbered in increments ranging from "Very Light" (disk #95) to "Very Dark" (disk #25) ( Pugash 1995 ) These classification disks are standards used by the Brazilian Coffee Industries Association (ABIC) ( Farah and others 2005 ; Farah and others 2006a ) and are readily available for visual comparison with coffee, but the literature lacks information about their numerical color values, such as those used in the red, green, blue (RGB) and L*, a*, b* color systems, which can be measured by different color measuring instruments. In addition, Agtron spectrophotometers have not become popular due to the ir limited use for other food products T he proprietary unit used by the manufacturer, calle d "Agtron", has no known correlation with other color units normally used such as L*, a*, and b* values from CIELab (International Comission on Illumination) or red, green and blue values from the RGB color space the former more frequently used by the sc ientific community. Hunterlab (Reston, Virginia, U.S.A.) spectrophotometers are used in production and in laboratory for inspection of different types of materials and finished products including most types of foods ( MacDougall 1988 ) Different from many spectrophotometers, some instrument models feature specular exclusion allowing color on surfaces with reflectance to be assessed. The M inolta Chroma Meter (Osaka, Japan) is a portable hand held colorimeter used for measuring the average color of a food

PAGE 75

75 sample area by providing controlled illumination, either average daylight illuminant C (6774 K color temperature) or average daylight illu minant D65 (6504 K color temperature). It is widely used in the field of agriculture and foods, including coffee where it has been used to determine the lightness of beans ( Schenker and others 2002 ; Baggenstoss and others 2007 ; Mahattanatawee and others 2007 ; Hernandez and others 2008b ) Computer machine vision (MV) systems use digital cameras acquiring images from a view area, which are then analyzed with special computer software ( Balaban and Odabasi 2006 ; Balaban 2008 ) MV has widely been used in inspection lines allowing accurate grading and inspection in many industries including food ( Bros nan and Sun 2002 ; Pedreschi and others 2008 ) The objectives of this study were (1) to measure the color of the Agtron SCAA R oast Color C lassification System and that of ten different coffee roasts with HunterLab and Minolta colorimeters and MV system under both regular and polarized light, and (2) to determine the correlation between the color of roasted coffee and those of the Agtron SCAA classification disks using the above instruments and sensory analysis. Materials and Methods Agtron SCAA Roast Color Classification System The Agtron RCCS was purchased from the Specialty Coffee Association of America (Long Beach, California, U.S.A.). The complete classification set is comprised of 8 color disks numbered from 25 through 95 in ten point increments ; each disk of 3 inches in diameter is covered with a protective clear polyethylene layer and contains a color d escription associated with the degree of roast for that color (Table 4 1). The classification set is used in the coffee industry and other businesses to be visually compare d to the color of roasted ground coffee.

PAGE 76

76 Coffee and Roaster Equipment Dry processed g reen arabica coffee ( Coffea arabica L. ) variety B ourbon from Ipanema Coffees (Alfenas M.G., Brazil) harve sted during the 2008 2009 season were pre sorted in bean sizes between 6.35 mm and 6.75 mm and packaged in 60 kg burlap bags. Upon receiving in Gainesville, FL, beans were transferred to 46 cm length x 76 cm height x 0.1 mm thickness polyethylene bags fill e d with nitrogen, and frozen at 18C. Beans were thawed for about 24 h ours and roasted in an Ambex YM 2 propane roaster coupled with the automation Profile DCQ Plus system (Clearwater, Florida, U.S.A.) comprised of a Programmable Logic Controller (PLC) an d a computer interface to control the time temperature profile in the roaster. Coffee samples in ten different degrees of roast (colors) were produced using the same time temperature profile, by only changing the final temperatures to 200C, 205C, 210C, 215C, 220C, 225C, 230C, 235C, 240C and 243C. Those final temperatures were selected to produce roasted coffees with colors ranging from very light to very dark roast, similar to the range of roast colors found in the Agtron RCCS Figure 4 1 is the graphic representation of the time temperature roasting profile as well as the final temperatures selected to obtain our coffee in this study Table 4 2 shows the list of roasted coffee with their respective roast numbers and final roasting temperature. Roasted coffee beans were divided into two parts: part I was used for immediate color analysis, and part II was stored at 18C in glass jars for the sensory analysis. Before the sensory analysis, color was again assessed to determine color changes during storage. Ground coffee was obtained by grinding the beans through a Rancilio

PAGE 77

77 Rocky IV coffee grinder (Milano, Italy) and using ground coffee that passed through a Color Analysis The color of four different spots of the Agtron RCCS disks and of the ground coffee samples was assessed using a Minolta handheld CR 200 Chroma Meter (Minolta Camera Co., Osaka, Japan), a HunterLab ColorQuest XE spectrophotometer (Hunter Associates Laboratory, Inc. Reston, Va., U.S.A.) and a machine vision system. The Minolta colorimeter was calibrated with a standard white plate (D65 illuminant, Y = 92.11 x = 0.3164, y = 0.3329). The HunterLab was calibrated with a standard white plate (D65 illumintant, X = 81.01, Y = 85.77, Z = 89.24) and was set to read reflectance with both specular included and specular excluded modes. L a and b values on both instruments were measured under D65 illuminant, which corresponds to a noon day light or 6504 K. The machine visio n system consisted of a light box ( Luzuriaga and others 1997 ) a Nikon D200 digital color camera, and a Nikon DX 18 200mm VR II Lens (Nikon Corp., Tokyo, Japan) connected to a computer through a USB cable, and Lens Eye Software (Engineering & Cybersolutions Inc., Gainesville, Florida, U.S.A.) developed in our lab, using Visual Basic for Windows (Microsoft, Redmond, Washington, U.S.A.). The light box used 2 fluorescent light bu lbs (Lumichrome F15W1XX, color temperature = 6500 K, color retention index=98, Lumiram, Larchmont, New York, U.S.A.) emulating the D65 illumination (natural daylight at noon). Diffuse light inside the box was obtained by using a Polycast acrylic nr 2447 pl astic sheet (Faulkner Plastics, Gainesville, Florida, U.S.A.) between the fluorescent bulbs and the sample space. Rosco 730011 polarizing sheets of 43.2 cm x 50.8 cm (Stamford, Connecticut, U.S.A.) were installed at the inside facing

PAGE 78

78 surface of the light b ox to create polarized lighting conditions, and a 72 mm Hoya circular polarizing Pro 1 glass filter (Tokyo, Japan) was mounted to the camera lens. The polarizing filter was set to an angle of 90 to the orientation of the polarized light to obtain maximum polarization (cross polarization), and of 0 for non polarization. The Nikon camera settings are listed in Table 4 3 Color analysis was performed using Lens Eye software. The Agtron RCCS disks were placed individually at the bottom of the light box and a picture was captured from the digital camera set on a tripod facing the disk at the bottom of the light box. Similarly the roasted coffees were filled into glass Petri dishes (87 mm in diameter and 12 mm in height) and placed individually at the bottom of the light box. The captured images (1000 x 669 pixels) taken with the MV system were calibrated with a Labsphere (North Sutton, New Hamshire, U.S.A.) standard red plate (L* = 48.62, a* = 49.04, b* = 25.72). The images were 24 bit color, meaning that in the Red, Green, Blue (RGB) color space, each color axis was represented by 8 bits or 28 = 256 different values. Using the LensEye software, first the RGB values of every pixel of a sample image were read, then this color information was converted to the L*, a *, and b* values, and averaged for each sample image. This resulted in the average L* values of 2 different samples using e quation 4 1 ( 4 1) Where x and y refer to the sample being tested and to the color anchor respectively.

PAGE 79

79 Sensory Analysis A sensory analysis was conducted to find a matching Agtron RCCS disk for each of the coffee roast samples tested. One day of testing was conducted by a panel comprised of students and staff from the University of Florida campus. Signs were used to obtain the panelists needed for the study. Compensation was provided to them for their time. The study was conducted at the Universit y of Florida and data was collected with a computer data center system (Compusense Five 3.6 Sensory Analysis Software for Window s, Compusense, Guelph, Canada). Panelists were screened for color blindness using Ishihara charts ( Hutchings 1999 ) to determine s uitable candidates for the color matching panel. A total of 3 Ishihara charts were used and candidates unable to read one or more charts were not considered for the sensory tests. Panelists were asked demographic questions regarding gender and age group. F ive coffee roasts R60, R65, R69, R71, and R77 were arbitrarily selected to represent the range of coffee color from very light to very dark roast. Ground samples of those coffee roasts were placed into 82 mm diameter Petri dishes (Fisher Scientific, Pittsb urgh, Pa., U.S.A.) and covered with black painted Petri lids to hide samples. Duplicates from the same first five samples were used for the second set of samples. Therefore p anelists were given a total of ten samples divided into two sets of five, both gro ups in randomized orders of presentation. They were asked to evaluate each set of samples separately by opening the lids of each sample at a time, and matching the color of that sample with one of the eight Agtron RCCS disks. Appendix B presents a copy of the screening and sensory tests used in this study.

PAGE 80

80 Illumination in the sensory room w as obtained with 3 sets of 4 fluorescent light bulbs from General Electric (Fairfileld, CT, U.S.A.) model F96T12/SP65, 75W, Daylight, 6500 Kelvin. Samples presented to th e panelists were carefully placed in such a way that no light reflection was obtained from the room illumination, or from the exterior through the windows. The lighting conditions at the location in the room where the panelists evaluated the samples were m easured with a light meter from International Light (Newburyport, MA, U.S.A.). Statistical Analysis All statistical analyses were performed in SAS version 9.3 (SAS Institute Inc., Cary, N.C., U.S.A.). Color data of the coffee color classification system an d of the ground coffee samples were analyzed using analysis of variance (ANOVA). Differences between values were determined by mean comparison using Tukey test ( = 0.05) Color matching between the sensory analysis and the instruments were analyzed using binomial regression Differences between instruments were determined by Least Square Means (LS Means). Results and Discussion Comparison of Color Measurement of the Agtron RCCS with Different Instruments The L*, a*, and b* values of the roast classificati on disks were measured with Minolta colorimeter (Mi), HunterLab with specular included (HLsi) and excluded (HLse), and machine vision under polarization (MVp) and non polarization (MVnp) (Figure 4 2). In all systems, L*, a* and b* values decreased as the r oasting degree increased, except for a* values between the first two classification disks. Significant differences (p < 0.05) were observed in L* values between classification disks read by a given instrument

PAGE 81

81 except for the MVnp. This latter instrument set the same instrument, except in the MVnp between disks 85, 75, 65, and 55 and also between disks 65, 55, 45, 35, and 25 in the MVnp. By far, the highest variation between two consecutive color classification disks was human eye, in a study about color quality of pepper powders it was demonstrated that samples, 0.5 to 1.5 a slight difference, 1.5 to 3.0 a just noticeable difference, above 3.0 a remarkable d ifference, and 6.0 and higher an extremely remarkable distinction in color ( Kim and others 2002 ) Another approach to compare instruments was to determine the magnitude of classification disks. Comparing the instruments used in this study, it can be observed that the MVp obtained signifi MVp system was more sensitive to changes in those color attributes. The average L*, a*, and b* values of the c lassification disks and those of the Labsphere standard red plate measured in all systems, together with the color

PAGE 82

82 representation of those values and the original image of the Agtron RCCS disks for visual comparison are presented in Table 4 4 When comparing each individual classification disk in all instruments, no significant differences between L* values were the other three instruments. The a* and b* values = 0.05) than the other systems for all disks except for the b* values of the two darkest disks T he MVp system seemed to produce bet ter color representation of the Agtron RCCS disks. Our assumption is that MVp was capable of minimizing the specular reflectance caused by the plastic coating present on the surface of the disks better than any other instrument. The color of the Labsphere red plate was very similar among instruments showing that probably for surfaces where reflectance is not a contributing factor, all systems would perform similarly. In a study comparing instruments in measuring the color of salmon filets, much better color representations was obtained using a MV system, in contrast with a Minolta colorimeter ( Yagiz and others 2009 ) In that study polarization technique was not applied Comparison of Color Measurement of Roasted Coffees U sing Diff erent Instruments The L*, a*, and b* values of ten ground roasted coffee and of ground green coffee the anchor color, are presented in Figure 4 3. L* values decreased wit h an increase in the final roasting temperature for all instruments, while a* and b* values initially

PAGE 83

83 increased and then decreased as the coffee roasts became darker, except in the HLsi system where no increase was observed for b* values. When comparing th and the darkest roasted coffee, significant differences were detected between all showing that this instrum ent is significantly more sensitive than the other instruments to color changes of roasted coffee. Table 4 5 shows the L*, a*, and b* values from the ground roasted coffees and the ground green coffee measured in all instruments, together with the color re presentation of those values, and the original image of the ground coffees. The comparison among instrument. When comparing each individual coffee sample in all instru ments, it was observed that all L*, a*, and b* values were significantly different in the MVp system than the values obtained in the other instruments. L* values were significantly higher in the green coffee but significantly lower in all of the roasted co ffees in this system. The visual comparison between the color representation and the actual color of the samples also shows a dramatic difference among instruments. Apparently all instruments but HLsi performed well represent ing the color of ground coffees HLsi obtained lighter colors as it can be observed from the higher L* values on the roasted samples. Correlation B etween Sensory Analysis and Instrumental Measurements The lighting conditions of the location in the room where the sensory panel took 2 which was later converted

PAGE 84

84 to 1100.04 lux, assuming the lighting spectrum average of 555 nm. According to the Illuminat ing Engineering Society of North America, the recommended illuminance levels for detailed tasks and for reading in the United States are in the range of 1000 to 2000 lux and 200 to 500 lux respectively ( Mills and Borg 1999 ) In this study, seventy eight panelists were screened for color blindness using considered for the sensory study. The number of male and female panelists divided by age groups is presented in Table 4 6 The color of the ground co ffee used in the sensory study was re assessed by MVp on the day of the sensory analysis and no significant changes were obtained compared Panelists were asked to match the c olor of different ground roasted coffees with one of the eight Agron/SCAA color classification disks. The cross tabulation of the sensory test is shown in Table 4 7 where the number of times the color of each classification disk was matched with the color of the various ground coffees is presented. Forty one panelists selected the color classification disk 95 as the disk with best matching color for coffee R60, the lightest roast, while 29 selected disk 85 and 2 panelists selected disk 75. When asked about panelists chose disk 95 and only 4 selected disk 85. No panelists picked disk 75. In both for coffees R69 and from 75 to 85. For the remaining darker coffee roasts R71, R77, and R65 and their replicates, some minor shifts in disk

PAGE 85

85 selection were observed but no changes in the most selected disks occurred when the pairs of duplicates were compared. The comparison b etween the color obtained from the instrumental readings and those from the sensory color matching was done in a two step process. The first step was to perform the c olor matching of the instruments by comparing the color difference between each ground roasted coffee and the classification disks. obtained was then used to determine the best matching classification disk for each of the roasted coffees. Table 4 8 is an example of how the instrumental color matching was performed to determine t he color of four coffee roasts. Finally the comparison between the instrumental and sensory color matching was done by cross matching the number of times the color of the coffee roast s assessed with instruments matched with the color chosen by the sensor y panelists. Since the color classification disks are meant to be used for visual matching with coffees, we assumed that the sensory analysis would be considered the reference method for color matching. A logistic or binomial regression analysis was carrie d out to determine the matching probability of each instrument in contrast with the sensory panel results, considering the binomial scores of 0 and 1 obtained for each un matching and matching colors, respectively. The cross matching of sensory analysis an d instruments is presented in Figure 4 4. Significant differences among instruments were observed ( = 0.05), while MVp was significantly better than the other instruments with 60.1% of matching probability. HLse

PAGE 86

86 obtained the second best cross matching sco re with 52.4%, significantly higher than HLsi, Mi, and MVnp. MVnp and Mi were not significantly different. Summary The MVp seemed to perform better than the other systems tested in measuring the color of ground roasted coffees and of the Agtron RCCS disks. Instruments not capable of excluding specular light, such as the MVnp, resulted in poor color representation of surfaces with high reflectance such as th ose from the classification disks; nevertheless they performed fairly well for matte surfaces, accepta bly representing the actual color of the coffee roasts. As a result, the cross matching of what MVnp defined as the color of the coffee roasts was dramatically low. T he systems HLse and MVp technically designed to exclude spec ular light, presented the bes t matching probability results, the latter significantly higher than the former Although the sensory color matching was used as reference to compare the instrumental performance, limitations from this method were detected. The variation in the results obt ained from the sensory analysis was noticeable. In all roasted coffee samples, an average of the most selected classification disk obtained was 64.7%. The remaining 35.3% o n average were distributed through the neighboring disks. Considering the score obta ined with MVp in the cross matching of 6 0 1 %, and the upper limit set by the sensory panel of 64.7%, we could conclu de that a corrected score for MVp would be 93%. That correction could be also applied to all instruments; however MVp would remain significa ntly better than the others. With these findings, we also concluded that m ore studies should be conducted in sensory for samples

PAGE 87

87 with high reflectance. The use of 2 sets of the same coffee samples in the sensory was an arbitrary decision and more studies should be conducted to try to understand the effect of training of panelists as they match the same sample more than one time. In addition, only 5 different coffee roasts were used, totaling ten samples with their duplicates, but greater number of different coffee roast samples may allow us to observe trends not seen under the conditions used in this study. If that is the case, other factors such as fatigue may play a role for panelists and a balance in number of samples versus panelist fatigue should be considered. Furthermore, the high variability obtained in the matching sensory test demonstrate d that probably sensory analysis is n ot the best reference measuring system. Further studies should be considered in correlating the measurements obtained from coffee roasts using an Agtron colorimeter and instruments such as the ones from this study. It can also be concluded from this study that MVp can be possibly used for other applications especially where specular reflectance is present and a method to obtain real colors is needed, such as in high surface moisture foods including meats, fish, and others. MV has been already demonstrated a s extremely useful for non uniformly colored foods. With the addition of excluding specular, MV would become more valuable for color assessment.

PAGE 88

88 Table 4 1. Agtron SCAA Roast Color Classification disk numbers and correspondent degree of roast descriptions Disk Degree of roast 95 very light 85 light 75 moderately light 65 light medium 55 medium 45 moderately dark 35 dark 25 very dark Table 4 2 List of c offee roast numbers and their final roasting temperatures. Roast number Final Roasting Temperature ( C) R66 200 R60 205 R61 210 R69 215 R70 220 R71 225 R80 230 R77 235 R78 240 R65 243 Roasts with an asterisk were arbitrarily selected for the sensory test. Table 4 3 Settings used in the Nikon D200 camera. Setting Non polarized Polarized Exposure Mode f/8 F/7.1 Shutter Speed (sec) 0.5 0.5 Aperture 0 EV* 0 EV* Exposure Compensation 0 EV* 0 EV* Sensitivity (ISO) 250 250 Color temperature (K) 5600 5000 Hue adjustment 6 6 *Exposure Value Since polarization reduces the total available light, aperture setting was adjusted to allow more light in the case of Polarized settings.

PAGE 89

89 Table 4 4. Average L*, a*, and b* values from Machine Vision, Minolta and HunterLab of the Agtron RCCS disks and those of stand ard red plate, and their color representation. Data represents the mean of n = 4. Values with similar capital letters in a row are not significant ly different (Tukey, p > 0.05). Comparisons of the same attribute within rows are presented in capitalized letters while comparisons within columns are not capitalized.

PAGE 90

90 Table 4 5 Average L*, a*, and b* values from machine vision, Minolta and HunterLab of green and roasted ground coffee, and their color representation. Data represents the mean of n = 4. Values with similar letters are not significantly different (Tukey, p > 0.05). Comparisons of the same attribute within rows are presented in capitalized letters while comparisons within columns are not capitalized.

PAGE 91

91 Table 4 6 Numbe r of panelists divided by age group in the sensory study. Age Group (years of age) Male Female Under 18 0 0 18 29 24 36 30 44 8 3 45 65 0 0 Over 65 1 0 Totals 33 39 Table 4 7 Sensory color matching of ground roasted coffee samples with the Agtron RCCS disks. Data represents the number of times each classification disk was selected per coffee roast sample by the panelists. Agtron RCCS disks Coffee Samples 95 85 75 65 55 45 35 25 Most selected disk R60 41 29 2 0 0 0 0 0 95 R69 1 17 42 12 0 0 0 0 75 R71 0 0 3 27 35 0 0 0 55 R77 0 0 0 0 18 53 1 0 45 R65 0 0 0 0 0 11 57 4 35 R60' 68 4 0 0 0 0 0 0 95 R69' 2 36 31 3 0 0 0 0 85 R71' 0 1 2 41 24 3 1 0 65 R77' 0 0 0 2 27 41 2 0 45 R65' 0 0 0 0 0 6 52 14 35 R oast code s followed by prime symbol s correspond to the duplicate d roast sample Coffee samples on the left column were ordered from low to high degree of roast. Most selected disk refers to the color classification disk with the highest number of selections by pan elist s

PAGE 92

92 Table 4 L* 27.9 24.4 22.1 17.3 13.5 6.2 4.7 3.2 a* 14.2 17.1 15.6 14.6 8.7 2.9 1.3 2.5 b* 24.0 24.1 19.9 15.5 7.0 1.8 1.5 1.5 Classification disk L* a* b* S** 95 85 75 65 55 45 35 25 L M ** 9.91 10.35 7.35 R80 24.8 23.1 18.3 11.8 4.0 10.0 15.5 17.0 4.0 55 6.7 6.08 3.47 R77 30.6 29.3 24.5 18.2 8.1 3.7 9.1 10.5 3.7 45 5.58 1.98 1.28 R78 34.1 33.2 28.4 22.4 11.9 1.2 4.4 5.8 1.2 45 4.67 1.07 0.31 R65 36.9 36.2 31.5 25.7 15.1 4.7 1.2 2.4 1.2 35 L*, a*, and b* values on the left are from the coffee samples and those on the header top from the classification disks. ** S and M are respectively the coffee roast sample and the best matching classification disk.

PAGE 93

93 Figure 4 1 Roasting profile used to produce roasted coffees in this study. The curve represents the temperature vs. time profile, while the represent the final roast temperatures for the ten different coffe e roasts. Final temperatures were arbitrarily selected at 200C, 205C, 210C, 215C, 220C, 225C, 230C, 235C, 240C, and 243C based on color of the final product. 0 50 100 150 200 250 300 0 100 200 300 400 500 600 700 Temperature ( C) time (sec)

PAGE 94

94 Figure 4 2 Agtron RCCS disks assessed using machine vision under non polarization ( ), under polarization ( ), HunterLab spectrophotometer with specular excluded ( ) and specular included ( ), and 0 10 20 30 40 50 60 70 95 85 75 65 55 45 35 25 L* value Disk -5 0 5 10 15 20 95 85 75 65 55 45 35 25 a* value Disk -5 0 5 10 15 20 25 30 95 85 75 65 55 45 35 25 b* value Disk 0 5 10 15 20 25 30 35 40 45 95 85 75 65 55 45 35 25 Disk

PAGE 95

95 Figure 4 3 using machine vision under non polarization ( ), under polarization ( ), HunterLab spectrophotometer with specular excluded ( ), and specular included ( Data represents the mean of n = 4. 0 10 20 30 40 50 60 70 80 GC R66 R60 R61 R69 R70 R71 R80 R77 R78 R65 L* value ground coffee -10 -5 0 5 10 15 20 25 GC R66 R60 R61 R69 R70 R71 R80 R77 R78 R65 a* value ground coffee -5 0 5 10 15 20 25 30 35 40 GC R66 R60 R61 R69 R70 R71 R80 R77 R78 R65 b* value ground coffee 0 10 20 30 40 50 60 70 80 GC R66 R60 R61 R69 R70 R71 R80 R77 R78 R65 ground coffee

PAGE 96

96 Figure 4 4 Cross matching of the color obtained from the sensory analysis and from the Data represents the mean of n = 40. Values with similar letters within the same color bar s are not significantly different (Tukey, p > 0.05). a d c b d 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% MVp MVnp HLsi HLse Mi Matching Probability Instrument

PAGE 97

97 CHAPTER 5 COFFEE AROMA: FORMAT ION AROMA ACTIVE SULFUR VOLATILES DURING COFFEE ROASTING Roasting is the single most important process to produce the characteristics of color and flavor in coffee. Roasting conditions have major impact on physical, chemical, and sensory properties of coffee. During this high temperature step, the pale green color of raw beans gives way to the various degrees of brown color found in roasted beans, and the green bean like smell of raw beans is transformed into the desirable but complex coffee aroma. The knowledge of coffee volatiles is very incomplete and many compounds are expected to be identified yet. Until very recently, a list of more th an 800 volatile compounds have been reported in roasted coffee, with only a small fraction possessing aroma activity ( Holscher and Steinhart 1992 ; Parliment and Stahl 1995 ; Grosch 1998 ; Mahattanatawee and others 2007 ) Many of these volatiles are sulfur and nitrogen containing compounds formed via Maillard and Strecker reactions, and have major impact in coffee aroma. The importance of sulfur volatiles in roasted coffee cannot be understated. A dramatic increase in the number and concentration of sulfur volatiles from raw to roasted beans has b een reported ( Schenker and others 2002 ; Mahattanatawee and others 2007 ) In a study using arabica coffee from Thailand, 14 sulfur peaks were found in green b eans, and more than 10 fold this number was found for light, medium and dark roasts, most of these volatiles with no aroma activity. Out of the 37 aroma active compounds detected, methanethiol, thiophene, dimethyl disulfide, dimethyl trisulfide, 3 methyl 2 butenthiol, 2 methyl 3 furanthiol, methional, 4 mercapto 4 methyl pentan 2 one, furfuryl methyl sulfide, 3 mercapto 3 methylbutyl formate, 2 acetyl 2 thiazoline and

PAGE 98

98 2 furfurylthiol were tentatively identified in roasted beans. The remaining 23 sulfur peak s associated with aroma activity could not be identified ( Mahattanatawee and others 2007 ) Although not all sulfur volatiles present aroma activity, many s ulfur containing compounds exhibit intense smelling properties due to their extremely low odor thresholds. Depending on their levels in beverages and foods, they contribute favorably to the aroma or to off flavor. 4 mercapto 4 methylpentan 2 one has been described as having a pleasant black currant bud odor, but at higher levels it gives an unpleasa nt cat urine odor ( Schneider and others 2003 ) Furfuryl mercaptan, also known as 2 Furfurylthiol or 2 furanmethanethiol, is the most regarded coffee aroma compound. It has been described by a numb er of authors as the best known sulfur containing coffee flavor compound ( Adams and others 2005 ) one of the critical flavors in coffee ( Parliment and Stahl 1995 ) a key coffee aroma compound ( Grosch 1998 ) and the outstanding odorant of the sulfur containing fraction of roasted coffee ( Grosch 2001 ) Together with 4 vinylguaiacol, some alkylpyrazines and furanones, 2 furfurylthiol is defined as one of the character impact odorants in roasted coffee ( Czerny and others 1999 ) 2 Furfurylthiol is one of the few odorants which can be described as roasted coffee. Some studies using model systems suggest that it may be f ormed from the thermal degradation of free or polymeric forms of pentoses (ribose and/or arabinose) and sulfur containing amino acids (cysteine and/or methionine) during roasting. Hexose sugars may be a source upon fragmentation ( Holscher and Steinhart 1992 ; Parliment and Stahl 1994 )

PAGE 99

99 Methional, the Strecker aldehyde of methionine, is the key compound of cooked potatoes, and can be detected in most thermally treated foods. Methional can also undergo degradation to form the more volatile methanethio l, which in lower levels has a pleasant aroma, although at high concentration it exhibits a putrid smell ( Holscher and Steinhart 1992 ) There are several methods to define the degree of roast in coffee, and these include color, weight loss water content, bean final temperature ( Sivetz 1991 ) formation of 2 methylpyrazine ( Hashim and Chaveron 1995 ) the ratio of 5 caffeoylquinic acid and caffeine ( Purdon and McCamey 1987 ) in addition to those real time methods typically used in small scale roasters such as smell and sound, which rely primarily on the operator experience ( Parliment and Stahl 1995 ) Among all these methods, color of ground beans is the most used and accepted indicator of the degree of roast for coffee ( Baggenstoss and others 2008b ) The nature of the chemical reactions th at take place produc ing color and flavor in coffee beans have very little correlation during roasting Studies have demonstrated that the aroma volatiles produced from coffee roasted to the same final color, but using different time temperature roasting co nditions, were significantly different ( Schenker and others 2002 ; Baggenstoss and others 2008b ) From these studies, it has been also suggested that bean final roasting temperature has no direct relationship to the degree of roast as suggested by other authors. The objective of this st udy was to determine the effect of roasting coffee to the same degree of roast (defined by color), using two dissimilar time temperature profiles in a commercial low scale horizontal drum roaster, on the evolution of total sulfur

PAGE 100

100 compounds and aroma active sulfur volatiles, including 2 furfurylthiol, methional, furfuryl methyl sulfide, dimethyl disulfide, thiophene, 3 methyl 2 butenthiol, dimethyl trisulfide, 4 mercapto 4 methyl pentan 2 one and 2 acetyl 2 thiazoline. Materials and Methods Coffee and Roaste r Equipment Dry processed green arabica coffee ( Coffea arabica L.) variety Bourbon from Ipanema Coffees (Alfenas M.G., Brazil) harvested during the 2008 2009 season were pre sorted in bean sizes between 6.35 mm and 6.75 mm and packaged in 60 kg burlap bags Upon receiving in Gainesville, FL, beans were transferred to 46 cm long x 76 cm high x 0.1 mm thick polyethylene bags filled with nitrogen, and frozen at 18C. The moisture content of the beans was 9.98% dry basis. Beans were thawed for about 24 hours and roasted in an Ambex YM 2 propane roaster coupled with the automation Profile DCQ Plus system (Clearwater, Florida, U.S.A.) comprised of a Programmable Logic Controller (PLC) and a computer interface to control the t ime temperature profile in the roaster. To create two reproducible and dissimilar time temperature roasting profiles, the heat output in the roaster interface was manually adjusted to 100% and 33% of the roaster heat power, and these were referred to as h igh temperature short time (HTST) and low temperature long time (LTLT) profiles respectively. Based on preliminary experiments, eight samples of different degree of roast were selected in the HTST profile, and their color were assessed to obtain the L*, a* and b* values. For the LTLT profile, another eight samples were selected by trial and error, comparing their color with the eight samples obtained in the former roasting profile, and selecting those of the same L* values.

PAGE 101

101 Coffee samples were ground using a Rancilio Rocky IV coffee grinder (Milano, Italy). Beans used in the color analysis were ground to a medium fine setting and passed through a 425 m sieve (Fisher Scientific, Pittsburgh, Pennsylvania, U.S.A.). Those samples intended for GC analysis were ground to a coarse setting and then This coarser setting was selected to avoid reduction in flow rate during the volatile exposure using the air sampler and trap system. Color Analysis The color of four different spots of the ground coffee samples was assessed using a machine vision system, consisting of a light box ( Luzuriaga and others 1997 ) a Nikon D200 digital color camera, and a Nikon DX 18 200mm VR II Lens (Nikon Corp., Tokyo, Japan) connected to a computer through a USB cable, and Lens Eye Software (Engineering & Cybersolutions Inc., Gainesville, Florida, U.S.A.) developed in our lab, using Visual Basic for Windows (Microsoft, Redmond, Washington, U.S.A.). The light box used 2 fluorescent light bulbs (Lumichrome F15W1XX, color temperature = 6500 K, color retention index=98, Lumiram, Larchmont, New York, U.S.A.) emulating the D65 illumi nation (natural daylight at noon). Diffuse light inside the box was obtained by using a Polycast acrylic nr 2447 plastic sheet (Faulkner Plastics, Gainesville, Florida, U.S.A.) between the fluorescent bulbs and the sample space. Rosco 730011 polarizing she ets of 43.2 cm x 50.8 cm (Stamford, Connecticut, U.S.A.) were installed at the inside facing surface of the light box to create polarized lighting conditions, and a 72mm Hoya circular polarizing Pro 1 glass filter (Tokyo, Japan) was mounted to the camera l ens. The polarizing filter was set to an angle of 90 to the orientation of the polarized light to obtain maximum polarization (cross polarization). The Nikon camera settings are listed

PAGE 102

102 in Table 5 1. Color analysis was performed using Lens Eye software. Sa mples of ground roasted coffee were placed into glass Petri dishes (87 mm in diameter and 12 mm in height) individually at the bottom of the light box and a picture was captured from the digital camera set on a tripod facing the sample at the bottom of the light box. Likewise, images of the Agtron SCAA Roast Color Classification System (Reno, NV, U.S.A.) disks were captured. The images (1000 x 669 pixels) taken with the MV system were calibrated with a Labsphere (North Sutton, New Hamshire, U.S.A.) standard red plate (L* = 48.62, a* = 49.04, b* = 25.72). The images were 24 bit color, meaning that in the Red, Green, Blue (RGB) color space, each color axis was represented by 8 bits or 28 = 256 different values. Using the LensEye software, first the RGB values of every pixel of a sample image were read, then this color information was converted to the L*, a*, and b* values, and averaged for each sample image. This resulted in the average L* a* b* color of each sample. Degree of roast comparison was based on the lightness value (L*) of the L*a*b* color space. Degree of Roast The degree of roast of coffee was determined using a method developed in our lab, by comparing the color of ground coffee samples to the Agtron SCAA Roast Color Classification System (Reno, NV U.S.A). The color difference E (5 1) was calculated between the color of each coffee sample and the eight roast classification disks. The classification disk that scored the least E was selected, and the degree of roast of the coffee sample was named a ccording to the Agtron SCAA scale. This method was fully described in chapter 4. (5 1 )

PAGE 103

103 X and Y refer to the roast classification disk being tested and to the coffee color respectively. Sulfur Volatile Analysis Dimethyl disulfide, dimethyl trisulfide, thiophene and 2 furfurylthiol were obtained from Acros Organics (Thermo Fisher Scientific, Pittsburgh, Penn., U.S.A.), 3 methyl 2 butenthiol and 4 mercapto 4 methyl pentan 2 one from Oxford Chemicals (Fruta rom, Hartlepool, U.K.), 2 methyl 3 furanthiol, 2 acetyl 2 thiazoline and methional from Sigma Aldrich (St. Louis, MO, U.S.A.), and furfuryl methyl sulfide from R.C. Treatt and Co. Ltd (Suffolk, U.K.). On day one, samples of 200 mg of coffee, ground immedia tely after roasting, were introduced between glass wool into a tube made from glass Pasteur pipets with the tapered end removed (Figure 5 1). The prepared tubes were then stored in glass jars at room temperature until the following morning for analysis. On day two, the tube with sample was connected to a Tenax GR trap (Gerstel, Germany) and a suction pump Gilian Low Flow Air Sampler (Sensidyne, Clearwater, Fla.) model LFS 113DC pulled air through the tube and trap for five minutes at constant flow. Samples were analyzed in triplicate within 24 hours of roasting. Flow rate of the sampling system was measured with a Brooks Instruments (Hartfield, Penn., U.S.A.) glass tube flow meter model number 1355FBH7AEA1A in standard cubic centimeters per minute (sccm) An average of 225 sccm was found for our system using a Tenax GR trap. Sulfur compounds were analyzed using a Pulsed Flame Photometric Detector (PFPD) in the square root mode (Model 5380, Ol Analytical Co., College Station, TX, U.S.A.) coupled to an Agilent 7890A gas chromatograph (Santa Clara, Cal., U.S.A.).

PAGE 104

104 Separation was accomplished using a polar DB wax capillary column (30 m x 0.32 mm i.d. x 0.5 m film thickness, J&W Scientific, Folsom, Cal., U.S.A.). The oven temperature was programmed from 60 180 C at 3 C/min followed by a 5min hold at 180 C. Helium was used as carrier gas at flow rate of 1.5 mL/min. Injection of volatiles was performed using a Thermal Desoption System (TDS, Gerstel, Germany) coupled to a Cooled Injection System (CIS 4, Gerstel, German y). The cryotrap in the CIS 4 was first cooled to 100 C with liquid nitrogen, then the trap was inserted into the desorption unit. The TDS was programmed to increase the temperature from 30 C to 240 C at a rate of 100 C/min. After the complete desorption cycle in the TDS, the temperature in the CIS 4 was programmed to increase at a rate of 10 C/sec until 240 C with an additional 30 sec at this latter temperature. The injector was operated in splitless mode at 200 C, and the detector at 250 C. Identificatio n and quantification of sulfur volatiles were tentatively confirmed by LRI comparison and by spiking with authentic standards. Odor Activity Value (OAV) was calculated using e quation 5 2 ( Rothe 1976 ) (5 2) Statistical Analysis All statistical analyses were performed in SAS version 9.3 (SAS Institute Inc., Cary, N.C., U.S.A.). Color data of the ground coffee samples and the concentration of sulfur volatiles were analyzed using analysis of variance (ANOV A). Differences between values were determined by mean comparison using Tukey test ( = 0.05). PCA was applied to the eight attributes ( sulfur compounds peak areas ) using Unscrambler X version 10.1 (Camo, Woodbridge, N.J., U.S.A) using correlation matrix.

PAGE 105

105 Results and Discussions Coffee Roast The graphical representation of the two time temperature roasting profiles showing the final temperatures selected to obtain our coffee samples in this study is presented ( Figure 5 2 ) The degree of roast between sample s of similar color were compared based on the lightness value (L*), from the L*a*b* color space, and no significant differences were found ( = 0.05). The use of L* value has been extensively evaluated in other studies to describe the degree of roast ( Purdon and McCamey 1987 ; Schenker and others 2002 ; Baggenstoss and others 2007 ; Baggenstoss and others 2008b ) Table 5 2 presents the final sampling temperature, L* values, and degree of roast for the coffee samples roasted under HTLT and LTLT profiles. GC Identifications and Quantifications A total of nine sulfur volatiles were separated, tentatively identified and quantified using GC PFPD based on the Linear Retention Index (LRI) in a polar DB wax column, authentic standards and spiked runs (Figure 5 3). The list of sulfur volatiles observed in this study and their peak areas are presented in Table 5 3 and in Appendix C respectively A general increasing trend in total sulfur volatiles was detected with the progress of roasting (Figure 5 4). However in the HTST profile, a significant decrease in total sulfu r was noticed in the final two sampling points. The reason for this decrease is unknown but may be due to changes in physical properties of the beans and thermal degradation. Volume of beans as well as their internal pores sizes increase significantly more in fast roasting processes ( Schenker and others 2000 ; Mwithiga and Jindal 2003 ) and the loss of volatiles could be favored under these conditions. Schenker and others

PAGE 106

106 ( 2002 ) also studied the effect of different roasting profiles on coffee flavor, using a spouted bed roaster, and suggested that the lower concentrati on of total aroma compounds found in HTST roasting process was due to more extensive thermal degradation. Chromatograms of light (L* = 25.8) and dark (L* = 4.2) roasts are presented in Figure 5 5 and Figure 5 6 respectively. These plots illustrate the diff erences in sulfur peaks between HTST and LTLT roasting profiles. The concentration of 2 furfurylthiol (FFT) increased in both HTST and LTLT roasting processes. However under HTST, a significantly higher concentration of this volatile was detected in the four darkest degree s of roast ( = 0.05) (Figure 5 7 ). The formation of FFT was very low in the first stages of roasting but increased more rapidly as the degree of roast increased. This trend was similar to other studies using arabica coffee ( Parliment and Stahl 1994 ; Schenker and others 2002 ) The activation energy (E a ) of FFT studied in cysteine pentose model systems was relatively high in contrast to other aroma compounds, and was de termined to be 48.6 Kcal/mole ( Parliment and Stahl 1994 ) which explains the lo w generation during the first roasting stages. In that model study, the authors also observed that the generation of FFT reached a maximum, then decreased, suggesting that FFT decomposes or reacts further to produce other products. A relatively wide range of FFT concentration has been reported in roasted coffee, depending on the coffee type, origin, post harvest treatments, roasting processes, and some other variables. Tressl ( 1989 ) reported that the overall concentration range of FFT in roasted coffee was between 500 and 4 000 g Kg 1 typically between 1 000 and 2 000 in Arabica, and between 2 000 and 3 800 g Kg 1 in robusta. Other reports

PAGE 107

107 include concentrations of 1 680 g Kg 1 in medium roast arabica beans ( Mayer and others 2000 ) 1 730 g Kg 1 in medium roast robusta ( Grosch 1998 ) and 1 150 g Kg 1 in Colombian wet processed arabica beans ( Poisson and others 2009 ) Using the extraction and GC technique of this study, we obtained near 1 000 g Kg 1 in dark r oast (L* = 4.2) for the HTST profile. However samples from lower degree s of roast were significantly lower than most reported values. Many reasons would explain a reduced recovered amount of FFT, including the current technique and methods, but one additio nal reason is the fact that the raw beans of this study were dry processed in the postharvest treatment. Since lower pH highly favors the generation of FFT, dry processed beans tend to yield lower FFT versus fermented coffee coming from wet postharvest pro cesses ( Parliment and Stahl 1994 ) The Odor Activity Values for FFT, presented in Figure 5 8 show how significant the presence of this sulfur volatile is in coffee flavor. The odor threshold of 0.01 g Kg 1 is relatively low, typical of most aroma active sulfur volatiles. The OAV is below 10,000 in the first four stages of roasting but increases rapidly in the HTST process. The OAV is significantly higher from light medium roasts (L* = 17.2), and a remarkable difference was detected for dark roast (L* = 4.2), peaking above 100,000 times the odor threshold of FFT. In general, LTLT pro file produced less FFT and therefore lower OAV than HTST. Significant differences were detected from light medium (L* = 17.2) to dark (L* = 4.2) roasts. The nutty like compound Furfuryl methyl sulfide (FMS) exhibited a similar generation trend as FFT in bo th roasting profiles (Figure 5 7 ). Likewise, the formation of

PAGE 108

108 FMS seems to be favored by high roasting temperatures, also suggesting a relative ly high E a for this compound. As for methional (MET), the generation trend was quite different than observed for FFT and FMS. Under HTST and LTLT conditions, MET peaked at medium (L* = 8.2) and at moderately dark (L* = 6.5) degree s of roasts respectively (Figure 5 7 ). Significant loss or possible degradation of MET was detected during the late r stages of HTST profil e ( = 0.05). The cooked potato like smelling compound that has been reported in raw and roasted coffee for many decades, was detected in relatively low concentrations in our study, below 100 g Kg 1 In an investigation using arabica coffee from various or igins, roasted to the same degree of roast, Grosch ( 1998 ) showed concentrations of MET in the range between 130 and 350 g Kg 1 using different extraction techniques. Lowest concentrations were found in beans from Ecuador and Brazil. As for its OAV, methional seems to have little impact on the overall coffee flavor (Figure 5 8 ). The odor threshold of 0.2 g Kg 1 is slightly higher than FFT, but its concentration in coffee is relatively low. Thiophene (TPN), dimethyl disulfide (DMDS) and 3 methyl 2 butenthiol (3M2BT) were not detected in green coffee nor in the first roasting stages (Figure 5 7 ). These compounds have a relative high vapor pressure (Table 5 3), meaning that possibly their loss due to volatility was higher than the other sulf ur volatiles in this study. Their presence w ere only detected from the light medium (L* = 17.2) degree of roast to dark roast (L* = 4.2) at relatively low concentrations. The presence of the sulfur like aroma of 3M2BT between 10 and 45 g Kg 1 is in accord ance with findings from other authors that reported its presence at 8.6 and 28 g Kg 1 in medium roasted coffees from

PAGE 109

109 Colombia ( Grosch 1998 ) However, the very low odor threshold of 3M2BT of 0.0003 g Kg 1 allowed this volatile to reach higher OAV than FFT (Figure 5 7 ). LTLT roasting profile produced significantly higher concentration of 3M2BT and consequently higher OAV at dark roast (L* = 4.2) than HTST The popcorn like, nutty like, or cooked rice like 2 acetyl 2 thiazoline was found in very low concentrations in coffee, below 5 g Kg 1 The compound showed a trend of formation and loss/degradation along both roasting profiles. 2 acetyl 2 thiazoline has the lowest vapor pressure among the sulfur volatiles studied (Table 5 3). In terms of odor activity, 2 acetyl 2 thiazoline threshold is found to be 1 g Kg 1 in water. The OAV was extremely low, since concentrations were just above this level. As for d imethyl trisulfide and 4 mercapto 4 methyl pentan 2 one these compounds co eluted under the techniques used in this study, and quantification was not possible. Peak area of the co eluted compounds is presented in Figure 5 7 Both compounds were not found in green coffee (L* = 66.5) but were detected immediatel y in the first intermediate coffee roast (L* = 53.1) and in all samples for both HTST and LTLT profiles, with a great er increase in peak area in medium roast (L* = 8.2). Dimethyl disulfide has been reported in concentrations of as low as 28 g Kg 1 ( Grosch 2001 ) and as high as 100 g Kg 1 ( Baggenstoss and others 2008b ) in roasted beans. The co eluted compound, 4 mercapto 4 methyl pentan 2 one has been identified in roasted coffee once, but without further quantification ( Mahattanatawee and others 2007 ) Principal Component Analysis (PCA) PC analysis was conducted to reduce the number of original variables (volatile concentrations in g/kg ) int o a fewer number of unobserved variables called principal components that are linear combinations of the original ones The main objective of

PAGE 110

110 PCA was to explain most of the variability of the volatile concentrations with the fewest of principal components and try to determine how HTST and LTLT a ffected the overall sulfur compound formation. PCA load plot (Figure 5 9) displays how the components (volatiles) fall relative to each other. All sulfur volatiles, but 2 furfurylthiol and furfuryl methyl sulfide are strongly correlated, as they are presented as a cluster of components near the center of the PCA load plot. PCA can also display where the samples fall relative to the variables most responsible for their differences Principal component 1 (PC1) explained 95% of the variation observed in the attribute intensity data, and principal component 2 (PC2) In this study, the PCA score plot showed which coffees were grouped w ith s imilar attributes (Figure 5 10 ). The cluttered center of the plot presents most of the samples that were taken at low roasting stages (light roasts), including raw beans. As the two different roasting profiles progress, a remarkable difference in trends wa s observed. Samples roasted under HTST grouped clearly in the upper quadrant, and samples roasted under LTLT grouped in the lower quadrant. HTST samples seem to be mostly driven by 2 furfurylthiol, and LTLT samples by furfuryl methyl sulfide, which can be confirmed by the significantly higher concentrations of these compounds in samples produced under the roasting profiles HTST and LTLT. Summary Roasting leads to dramatic increase s in sulfur volatiles in coffee. Analysis of aroma active sulfur compounds demonstrated that time temperature roasting conditions significantly affect their final concentrations, by changing their rate of formation and

PAGE 111

111 degradation, and/or by increasing thei r loss. Sulfur volatiles are known to produce many off flavors, though pleasant aromas are also caused by many sulfur containing compounds when at the right concentrations. Total sulfur compounds w ere significantly higher in moderately dark and dark coffee roasts when a LTLT roasting process was applied in contrast with a HTST roasting process for the same final degree of roast. However when individually analyzed, some compounds such as 2 furfurylthiol were produced in significantly higher concentrations i n the HTST profile. Seven compounds with aroma activity were tentatively identified and quantified based on LRI values using a polar column (DB wax), and compared to standards and standards spiked samples 2 furfurylthiol, furfuryl methyl sulfide and methi onal were detected in raw beans as well as in all coffee roasts. Four sulfur compounds, thiophene, dimethyl disulfide, 3 methyl 2 butenthiol, and 2 acetyl thiazoline were only detected in roasted coffees. Two other compounds with aroma activity, dimethyl t risulfide and 4 mercapto 4 methyl 2 pentanone, were not quantified due to co elution. PCA indicated that coffee roasted to the same color presented greater sulfur profile differences at higher degree of roasts (lower L* values). Raw beans and coffee of ver y low degree of roast showed great correlation, which was confirmed by the low sulfur compounds formed during the initial roasting stages. Considering the large number of aroma active compounds present in roasted coffee, an optimization of the roasting pro cess to yield better coffee aroma would require more extensive knowledge of all these volatiles. However, knowing that a few

PAGE 112

112 compounds have a major contribution to the final coffee aroma, an optimized roasting profile to enhance the formation of these vola tiles, which include the highly regarded 2 furfurylthiol, might be possible. The technique applied to this study, consisted of volatile trapping, followed by desorption using a cooled injection and a thermal desorption systems, showed promising results, co mparable to other lengthy methodologies. Further studies should include a comprehensive comparison of this technique with Solid Phase Micro Extraction (SPME), and solvent extraction, which were used in previous studies of coffee volatiles. In addition, stu dies involving brewed coffee would be extremely appropriate, as it is uncertain how much of the volatiles would be extracted in hot water.

PAGE 113

113 Table 5 1 Settings used in the Nikon D200 camera under polarized lighting Setting value Exposure Mode F/7.1 Shutter Speed (sec) 0.5 Aperture 0 EV* Exposure Compensation 0 EV* Sensitivity (ISO) 250 Color temperature (K) 5000 Hue adjustment 6 *Exposure Value Table 5 2 Final sampling temperature, L* values, average L* values, and degree of roast description of coffee roasted under HTLT and LTLT profiles. HTST LTLT Sampling Temp (C) L* Sampling Temp (C) L* Mean L* Degree of roast ** raw bean 66.5 0.08 66.5 Green coffee 180 52.7 0.69 178 53.6 0.28 53.1 Intermediate 1 190 44.6 0.2 186 45.2 0.65 44.9 Intermediate 2 200 35.5 0.21 197 35.8 0.74 35.6 very light 210 25.6 0.35 207 26 0.87 25.8 light 220 17.5 0.27 218 17 0.34 17.2 light medium 230 8.3 0.25 235 8.2 0.66 8.2 medium 235 6.6 0.21 240 6.4 0.38 6.5 moderately dark 243 4.1 0.11 249 4.4 0.27 4.2 dark ** Degree of roast is a standard rating for coffee roasting ( Pugash 1995 )

PAGE 114

114 Table 5 3. List of sulfur volatiles analyzed in this study. Nr Name LRI Wax Odor threshold ( g/kg) Vapor Pressure (mmHg)* Boiling Point at atm (C) 1 thiophene 1051 n.a. 40* 84 2 Dimethyl Disulfide 1102 12 1 28.5 110 3 3 methyl 2 butenthiol 1132 0.0003 2 14.4 128 4 4 mercapto 4 methyl pentan 2 one 1399 0.0008 3 0.843 n.a. 5 Dimethyl trisulfide 1400 0.01 1 1.07 165 6 2 Furfurylthiol 1450 0.01 4 3.98 154 7 Methional 1473 0.2 4 1.64 165 8 Furfuryl methyl sulfide 1504 n.a. 1.58 64** 9 2 acetyl 2 thiazoline 1776 1 5 0.0942 222 Odor threshold value reported by: 1 ( Buttery and others 1976 ) 2 ( Holscher and others 1992 ) 3 ( Schneider and others 2003 ) 4 ( Semmelroch and o thers 1995 ) 5, ( Guth and Grosch 1994 ) measured or estimated at 25 C ** measured or estimated at 12.5 C *** at 15 mm Hg n.a. data not available

PAGE 115

115 Figure 5 1. Drawings of (1) the original Pasteur glass pipet showing the cutting section, (2) the sample holder and (3) the complete sampling system. Figure 5 2 Roasting profile s of HTST (solid line) and LTLT (dashed line) used to produce the roasted coffees in this study. Sampling stages selected for our study are marked with O and along with their temperatures 178 186 197 207 218 235 240 249 180 190 200 210 220 230 235 243 0 50 100 150 200 250 300 0 100 200 300 400 500 600 700 800 900 Temperature ( C) Time (sec)

PAGE 116

116 Figure 5 3. Superposed chromatograms of two GC runs showing the spiked 2 furfurylthiol. Figure 5 4. Evolution of total sulfur compounds in coffee during LTLT (dashed line) and HTST (solid line) roasting profiles: some peaks saturated detector and actual concentration will be slightly higher than presented. * * * 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 Total peak area (thousands) L* value

PAGE 117

117 Figure 5 5 Comparison between chromatogram s of light coffee roast produced under HTST and LTLT roasting profiles. Compounds 1 4 are respectively: 3 methyl 2 butenthiol, 2 furfurylthiol, methional, and furfuryl methyl sulfide. Figure 5 6 Comparison between chromatograms of dark coffee roasts pr oduced under HTST and LTLT roasting profiles. Compounds 1 6 are respectively: thiophene, dimethyl disulfide, 3 methyl 2 butenthiol, 2 furfurylthiol, methional, and furfuryl methyl sulfide.

PAGE 118

118 Figure 5 7 Evolution of sulfur compounds for coffee d uring the LTLT (dashed line) and HTST (solid line) roasting profiles. 0 200 400 600 800 1,000 1,200 1,400 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 1 L* value 2 Furfurylthiol 0 100 200 300 400 500 600 700 800 0.0 20.0 40.0 60.0 80.0 1 L* value Furfuryl Methyl Sulfide 0 10 20 30 40 50 60 70 80 90 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 1 L* value Methional 0 20 40 60 80 100 120 140 160 180 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 1 L* value Thiophene

PAGE 119

119 Figure 5 7 Continued 0 5 10 15 20 25 30 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 1 L* value Dimethyl disulfide 0 10 20 30 40 50 60 70 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 1 L* value 3 methyl 2 butenthiol 0 1 2 3 4 5 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 1 L* value 2 acetyl 2 thiazoline 0 50,000 100,000 150,000 200,000 250,000 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 Peak Area L* value Dimethyl trisulfide and 4 mercapto 4 methyl pentan 2 one

PAGE 120

120 Figure 5 8 Evolution of sulfur compounds as OAV for coffee during the LTLT (dashed line) and HTST (solid line) roasting profiles. 0 20,000 40,000 60,000 80,000 100,000 120,000 140,000 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 OAV L* value 2 furfurylthiol 0 50 100 150 200 250 300 350 400 450 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 OAV L* value methional 0 1 2 3 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 OAV L* value Dimethyl disulfide 0 50,000 100,000 150,000 200,000 250,000 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 OAV L* value 3 methyl 2 butenthiol

PAGE 121

121 Figure 5 8 Continued Figure 5 9 Loading plot showing principal component 1 ( PC 1) and principal component 2 ( PC 2) for the sulfur volatiles studied. 0 1 2 3 4 5 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 OAV L* value 2 acetyl 2 thiazoline

PAGE 122

122 Figure 5 10 PCA score plot showing Principal Component 1 ( PC 1) and Principal Component 2 ( PC 2) for the sulfur volatiles studied. Numbers indicate the final temperature of the samples. Samples treated in HTST are presented in red color; those treated in LTLT in blue, and raw beans in green.

PAGE 123

123 CHAPTER 6 SUMMARY AND CONCLUSIONS Finding s from this stu dy will make color prediction possible during roasting in real time. Change in lightness ( L* values) during whole bean roasting was well predicted using 2 nd order reaction kinetics The high energy of activation of 145.8 KJ mol 1 is in accordance to other studies showing that E a for browning in foods are in the range of 37.7 and 167.5 kJmol 1 The rate constant k 0 was 2.1x10 12 L 1 sec 1 and followed Arrhenius relationship. Changes in a* and b* values, for ground green coffee roast ing, seem to follow irreversible reactions in series. The model showed good agreement with experimental data under isothermal conditions. However m ore studies at higher temperatures to obtain kinetic parameters seem necessary. Color conversion from L*, a*, and b* values to Agtron, can be accomplished in different instruments. The machine vision system under polarizing lighting seem ed to perform better than any of the other system s tested in measuring the color of ground roasted coffees and the Agtron SCAA R oast C olor C lassification System Although sensory color matching was used as reference to compare the instrumental performance, limitations from this method were detected. T he high variability obtained in matching the sensory test demonstrated that sensor y analysis was likely not the best reference measuring system. Further studies should be considered in correlating the measurements obtained from coffee roasts using an Agtron colorimeter and i nstruments such as the ones in this study. It can also be concl uded that MVp can be used in other applications especially where specular reflectance is present and a method t o obtain real colors is needed.

PAGE 124

124 There is evidence that roasting profile greatly affect s sulfur compound formation, but more work must be done to determine the best sulfur volatile balance that will lead to coffee with better aromas and that should include sensory analysis with professional coffee tasters Roasting leads to dramatic increases in sulfur volatiles in coffee. Analysis of aroma active sulfur compounds demonstrated that time temperature roasting conditions significantly affect their final concentrations Total sulfur compounds w ere significantly higher in moderately dark and dark coffee roasts when a LTLT roasting process was applied in contrast to a HTST roasting process for the same final degree of roast. However it was found in earlier study that the majority of sulfur volatiles were not aroma active at typical concentrations. W hen individually analyzed, some compounds were produce d in significantly higher concentrations in the HTST profile. Considering that just a few compounds have a major contribution to the final coffee aroma, an optimized roasting profile to enhance the formation of these volatiles, which include the highly reg arded 2 furfurylthiol, might be possible.

PAGE 125

125 APPENDIX A ROASTER AND COFFEE P ICTURES A B C D Figure A 1. Pictures of the drum roaster (A), front section showing the sampling system (B), cooling of roasted coffee samples (C), ProfilePlusDCQ automation system main screen (D).

PAGE 126

126 APPENDIX B SENSORY COLOR MATCHING Screening Test Today's Sample: C o f f e e To start the test, click on the Continue button below: Panelist Code: ________________________ Panelist Name: ________________________________________________ Thank you for agreeing to participate in this study. In the next screen(s) you will be presented with images showing some numbers in color. You will be asked to look at the image and type in the number y ou see in the screen to follow. Please press 'continue' button below

PAGE 127

127 Question # 1. Please type in the number you saw in the previous screen. You can go back to view it again by clicking 'Review Instructions' below.

PAGE 128

128 Question # 2. Please type in the number you saw in the previous screen. You can go back to view it again by clicking 'Review Instructions' below.

PAGE 129

129 Question # 3. Please type in the number you saw in the previous screen. You can go back to view it again by clicking 'Review Instructi ons' below.

PAGE 130

130 Please hit 'Next Sample' below and move over to the next computer station for the test. The test has ended. DO NOT click continue on the next screen. Thank you.

PAGE 131

131 Questionaire Today's Sample: C o f f e e To start the test, click on the Continue button below: Panelist Code: ________________________ Panelist Name: ________________________________________________ Thank you for agreeing to participate in this study. You are going to receive some ground coffee samples and 8 color disks. You will be asked to match the color of the coffee samples against the 8 reference disks. Please press 'continue' button below Question # 1. Please indicate your gender. Male Female

PAGE 132

132 Question # 2. Please indicate your age range. Under 18 18 29 30 44 45 65 Over 65

PAGE 133

133 Question # 3. Please indicate your age range. Under 18 18 29 30 44 45 65 Over 65

PAGE 134

134 Question # 4 Sample ______ Please choose a color disk that best matches coffee sample %01. 95 85 75 65 55 45 35 25

PAGE 135

135 The test has ended. DO NOT click continue on the next screen. Please lift your window to let the server know you have finished. Thank you.

PAGE 136

136 APPENDIX C SULFUR COMPOUNDS PEAK AREAS

PAGE 137

137 Table C 1. Data for s ulfur compounds found in coffee experiment (Chapter 5). (1) thiophene, (2) Dimethyl disulfide, (3) 3 methyl 2 butenthiol, (4) dimethyl trisulfide and 4 mercapto 4 methyl pentan 2 one, (5) 2 furfurylthiol, (6) methional, (7) furfuryl methyl sulfide, (8 ) 2 acetyl 2 thiazoline. L* values 66.5 53.1 44.9 Nr LRI Roast peak area H V peak area H V peak area H V 1 1039 LTLT nd N d N d HTST N d N d N d 2 1094 LTLT nd N d N d HTST nd N d N d 3 1128 LTLT nd N d N d HTST nd N d N d 4 1399 LTLT nd 20587 2581 D a 32231 16937 D a HTST nd 20222 6839 D a 14885 1950 D a 5 1450 LTLT 70136 39320 E a 14447 3374 E a 50112 21364 E b HTST 7013639320 C a 72994 30575 C a 202408 34164 C a 6 1473 LTLT 18000 10678 D a 48166 3959 D a 23879 6290 D a HTST 1800010678 D a 52083 27836 D a 62743 25100 D a 7 1503 LTLT 28114895044 E a 147418 59660 E a 153511 119137 E b HTST 281148 95044 G a 326102 100613 F G a 368476 56583 F G a 8 1776 LTLT nd 28851 4463 B a 34876 5465 B a HTST nd 19786 12855 A B C a 29304 4438 A B a Data represents the mean of n = 3 Values wi th similar capital letters ( H) are not significant different (Tukey, p > 0.05) in same row Values with similar letters (V) are not significant between concentrations of the same compound with the same L* value nd = not detected.

PAGE 138

138 Table C 1. Continued L* values 35.6 25.8 17.2 Nr Roast peak area H V peak area H V peak area H V 1 LTLT nd nd nd HTST nd nd nd 2 LTLT nd nd nd HTST 6037 0 C nd nd 3 LTLT nd nd nd HTST nd nd nd 4 LTLT 13104 1465 D a 23964 4578 D a 46746 4000 C D a HTST 15304 3147 D a 32762 4475 D a 27635 8003 D b 5 LTLT 134040 63282 E b 433235 105896 D a 801959 263598 C b HTST 307015 65024 C a 549588 100532 C a 2471424 102397 B a 6 LTLT 33751 21718 D b 240476 51746 C a 331842 59918 B a HTST 89921 10274 D a 271583 62623 C a 473355 71282 B a 7 LTLT 324193 68051 E b 819008 301131 D a 1284663 58632 C a HTST 557149 67669 E F a 753889 15108 E a 1316531 131000 D a 8 LTLT 20048 1673 B b 66494 3753 A 60901 8434 A a HTST 33909 6484 A a nd 41888 5533 A a Data represents the mean of n = 3 Values wi th similar capital letters (H) are not significant different (Tukey, p > 0.05) in same row. Values with similar letters (V) are not significant between concentrations of the same compound with the same L* value. nd = not detected.

PAGE 139

139 Table C 1. Continued L* values 8.2 6.5 4.2 Nr Roast peak area H V peak area H V peak area H V 1 LTLT 46960 6107 B a 55884 6569 B a 99905 25169 A a HTST 36745 2950 A B a 19791 4632 B b 39299 14567 A b 2 LTLT nd 23127 5254 B a 42910 2149 A a HTST 27559 5633 B 43916 16051 A B a 50670 7010 A a 3 LTLT nd 33977 14052 A a 57294 12566 A a HTST 14037 10323 A 11334 16028 A a 28491 3784 A b 4 LTLT 89019 18095 B C b 95081 56683 B a 211187 10361 A a HTST 144111 14455 A a 66333 22669 C a 100190 20392 B b 5 LTLT 1466102 42372 B b 1265406 153218 B b 2058333 250400 A b HTST 3119841 170375 B a 2377546 288909 B a 8563400 1410853 A a 6 LTLT 339415 69524 B b 456093 86156 A a 399570 32884 A B a HTST 595800 85626 A a 284568 93944 C a 250175 79816 C a 7 LTLT 3644234 331222 B a 3755426 117201 B a 5075070 344780 A b HTST 2378207 224594 C b 2862876 110353 B b 7942444 300316 A a 8 LTLT 31067 5168 B a 18964 11866 B a 54061 17387 A a HTST 37180 14903 A a 10206 14432 B C a 4692 6634 C b Data represents the mean of n = 3 Values wi th similar capital letters (H) are not significant different (Tukey, p > 0.05) in same row. Values with similar letters (V) are not significant between concentrations of the same compound with the same L* value. nd = not detected.

PAGE 140

140 APPENDIX D TIME TEMPERATURE TREATMEN TS AND COFFEE ROAST COLOR V ALUES Table D 1. Time temperature of the roasting treatments used to obtain coffee samples in an Ambex Roaster (Chapter 3). Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 0 228.9 229.9 231.2 230.0 1 226.2 224.7 228.4 226.4 2 220.8 219.8 222.7 221.1 3 215.4 214.9 217.7 216.0 4 210.4 210.3 212.7 211.1 5 205.7 205.8 208.1 206.5 6 201.2 201.4 203.8 202.1 7 196.8 197.2 199.4 197.8 8 192.8 193.1 195.1 193.7 9 188.8 189.1 191.2 189.7 10 185.0 185.4 187.3 185.9 11 181.4 181.8 183.7 182.3 12 177.8 178.3 180.1 178.7 13 174.4 174.9 176.4 175.3 14 171.2 171.7 173.0 172.0 15 168.0 168.6 169.7 168.8 16 165.0 165.6 166.4 165.7 17 161.9 162.4 163.2 162.5 18 159.1 159.5 160.2 159.6 19 156.3 156.8 157.5 156.9 20 153.6 154.3 154.7 154.2 21 150.8 151.7 152.1 151.6 22 148.3 149.3 149.4 149.0 23 145.7 146.9 147.0 146.6 24 143.3 144.6 144.6 144.1 25 140.9 142.3 142.2 141.8 26 138.7 140.2 140.0 139.6 27 136.4 138.1 137.8 137.4 28 134.4 136.0 135.6 135.4 29 132.4 134.1 133.7 133.4 30 130.5 132.2 131.7 131.4 31 128.6 130.3 129.7 129.5 32 126.7 128.7 128.0 127.8 33 124.9 127.0 126.2 126.1 34 123.2 125.3 124.6 124.4 35 121.6 123.8 123.0 122.8 36 120.1 122.3 121.4 121.3 37 118.6 120.8 120.1 119.8 38 117.2 119.5 118.6 118.4

PAGE 141

141 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 39 115.8 118.2 117.4 117.2 40 114.6 116.9 116.1 115.9 41 113.3 115.8 114.8 114.6 42 112.2 114.6 113.6 113.5 43 111.1 113.5 112.4 112.3 44 110.0 112.6 111.3 111.3 45 109.0 111.6 110.4 110.3 46 108.1 110.7 109.4 109.4 47 107.1 109.8 108.4 108.4 48 106.2 108.9 107.6 107.6 49 105.3 108.1 106.8 106.7 50 104.6 107.4 105.9 105.9 51 103.7 106.6 105.2 105.2 52 102.9 105.8 104.4 104.4 53 102.4 105.1 103.7 103.7 54 101.7 104.6 103.1 103.1 55 101.1 103.9 102.7 102.6 56 100.5 103.3 102.1 102.0 57 100.1 102.9 101.6 101.5 58 99.4 102.6 101.0 101.0 59 99.1 102.1 100.5 100.6 60 98.6 101.8 100.1 100.1 61 98.2 101.3 99.7 99.7 62 97.8 101.0 99.3 99.4 63 97.4 100.7 98.9 99.0 64 97.1 100.4 98.7 98.7 65 96.7 100.1 98.4 98.4 66 96.4 99.9 98.2 98.2 67 96.1 99.7 98.0 97.9 68 95.9 99.6 97.8 97.8 69 95.7 99.4 97.6 97.6 70 95.6 99.3 97.4 97.4 71 95.4 99.2 97.3 97.3 72 95.3 99.1 97.1 97.2 73 95.2 98.9 97.1 97.1 74 95.2 98.9 97.0 97.1 75 95.2 98.9 97.0 97.1 76 95.2 98.9 96.9 97.0 77 95.2 98.9 96.9 97.0 78 95.2 98.9 96.9 97.0 79 95.2 98.9 96.9 97.0 80 95.2 99.1 97.0 97.1 81 95.2 99.2 97.1 97.2

PAGE 142

142 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 82 95.3 99.3 97.1 97.2 83 95.4 99.3 97.1 97.3 84 95.4 99.3 97.2 97.3 85 95.6 99.4 97.3 97.4 86 95.7 99.7 97.4 97.6 87 95.8 99.7 97.6 97.7 88 95.9 99.8 97.7 97.8 89 96.1 100.1 97.8 98.0 90 96.2 100.1 98.0 98.1 91 96.3 100.3 98.1 98.2 92 96.5 100.5 98.3 98.4 93 96.7 100.7 98.4 98.6 94 96.9 100.8 98.7 98.8 95 97.1 101.0 98.8 99.0 96 97.3 101.2 99.1 99.2 97 97.5 101.5 99.3 99.4 98 97.7 101.8 99.4 99.6 99 97.9 102.0 99.7 99.9 100 98.1 102.3 99.9 100.1 101 98.3 102.6 100.1 100.3 102 98.6 102.8 100.4 100.6 103 98.8 103.1 100.6 100.8 104 99.1 103.3 100.9 101.1 105 99.3 103.4 101.1 101.3 106 99.6 103.7 101.4 101.6 107 99.7 104.1 101.7 101.8 108 100.1 104.3 102.0 102.1 109 100.2 104.6 102.3 102.4 110 100.5 104.9 102.6 102.7 111 100.8 105.2 102.8 102.9 112 101.0 105.5 103.1 103.2 113 101.3 105.7 103.3 103.5 114 101.6 106.1 103.7 103.8 115 101.9 106.4 103.9 104.1 116 102.2 106.8 104.2 104.4 117 102.5 107.0 104.6 104.7 118 102.7 107.3 104.7 104.9 119 102.9 107.6 105.0 105.1 120 103.2 107.8 105.3 105.4 121 103.6 108.2 105.6 105.8 122 103.7 108.6 105.9 106.1 123 104.1 108.8 106.3 106.4 124 104.4 109.1 106.6 106.7

PAGE 143

143 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 125 104.7 109.4 106.8 107.0 126 104.9 109.7 107.2 107.3 127 105.2 110.0 107.4 107.5 128 105.6 110.3 107.8 107.9 129 105.9 110.6 108.1 108.2 130 106.3 110.9 108.3 108.5 131 106.5 111.2 108.6 108.8 132 106.8 111.4 108.9 109.1 133 107.1 111.8 109.2 109.4 134 107.4 111.9 109.4 109.6 135 107.7 112.3 109.8 109.9 136 108.1 112.7 110.1 110.3 137 108.3 112.9 110.5 110.6 138 108.6 113.2 110.7 110.8 139 108.9 113.6 111.0 111.2 140 109.2 113.9 111.3 111.5 141 109.4 114.1 111.6 111.7 142 109.7 114.4 111.8 112.0 143 110.0 114.7 112.2 112.3 144 110.3 115.0 112.4 112.6 145 110.6 115.3 112.7 112.9 146 110.9 115.6 113.0 113.1 147 111.1 115.8 113.3 113.4 148 111.4 116.2 113.6 113.7 149 111.7 116.3 113.9 114.0 150 111.9 116.7 114.1 114.2 151 112.2 116.9 114.4 114.5 152 112.6 117.2 114.7 114.8 153 112.8 117.5 115.0 115.1 154 113.0 117.8 115.3 115.4 155 113.3 118.1 115.6 115.7 156 113.6 118.4 115.8 115.9 157 113.9 118.6 116.1 116.2 158 114.2 118.9 116.3 116.5 159 114.5 119.2 116.7 116.8 160 114.7 119.5 116.9 117.0 161 115.1 119.8 117.2 117.4 162 115.4 120.1 117.5 117.6 163 115.7 120.3 117.7 117.9 164 115.9 120.6 118.0 118.2 165 116.2 120.8 118.2 118.4 166 116.4 121.1 118.5 118.7 167 116.7 121.3 118.7 118.9

PAGE 144

144 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 168 116.9 121.6 118.9 119.1 169 117.2 121.9 119.2 119.4 170 117.4 122.2 119.5 119.7 171 117.7 122.4 119.8 120.0 172 117.9 122.7 120.1 120.2 173 118.2 123.0 120.3 120.5 174 118.4 123.2 120.4 120.7 175 118.6 123.4 120.7 120.9 176 118.8 123.7 120.9 121.1 177 119.1 123.9 121.2 121.4 178 119.4 124.2 121.4 121.7 179 119.6 124.3 121.7 121.9 180 119.9 124.7 122.0 122.2 181 120.1 124.9 122.3 122.4 182 120.3 125.2 122.5 122.7 183 120.6 125.4 122.8 122.9 184 120.8 125.6 123.0 123.1 185 121.1 125.9 123.3 123.4 186 121.2 126.1 123.5 123.6 187 121.4 126.3 123.7 123.8 188 121.7 126.5 123.9 124.1 189 122.0 126.8 124.3 124.4 190 122.3 127.0 124.3 124.5 191 122.4 127.3 124.7 124.8 192 122.7 127.5 124.9 125.1 193 122.9 127.7 125.2 125.2 194 123.1 127.9 125.3 125.4 195 123.3 128.1 125.6 125.6 196 123.6 128.4 125.8 125.9 197 123.8 128.6 126.0 126.1 198 124.1 128.8 126.2 126.4 199 124.3 128.9 126.5 126.6 200 124.6 129.2 126.8 126.8 201 124.7 129.4 127.0 127.0 202 124.9 129.6 127.2 127.2 203 125.2 129.8 127.4 127.5 204 125.4 130.0 127.6 127.7 205 125.6 130.2 127.9 127.9 206 125.7 130.4 128.1 128.1 207 126.0 130.6 128.3 128.3 208 126.2 130.8 128.6 128.5 209 126.4 131.0 128.8 128.7 210 126.6 131.2 128.9 128.9

PAGE 145

145 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 211 126.8 131.4 129.2 129.1 212 127.0 131.6 129.3 129.3 213 127.3 131.8 129.6 129.5 214 127.4 132.0 129.7 129.7 215 127.6 132.2 129.9 129.9 216 127.9 132.4 130.1 130.1 217 128.0 132.4 130.3 130.3 218 128.3 132.7 130.5 130.5 219 128.4 132.9 130.7 130.7 220 128.6 133.2 130.9 130.9 221 128.8 133.3 131.2 131.1 222 129.1 133.6 131.4 131.3 223 129.2 133.7 131.6 131.5 224 129.4 133.9 131.7 131.7 225 129.6 134.1 132.0 131.9 226 129.7 134.3 132.1 132.0 227 129.9 134.4 132.4 132.3 228 130.1 134.6 132.6 132.4 229 130.2 134.8 132.8 132.6 230 130.4 135.0 132.8 132.7 231 130.6 135.2 133.2 133.0 232 130.8 135.3 133.3 133.1 233 131.0 135.5 133.4 133.3 234 131.2 135.6 133.6 133.5 235 131.3 135.7 133.7 133.6 236 131.5 135.9 133.9 133.8 237 131.7 136.1 134.1 133.9 238 131.9 136.3 134.3 134.2 239 132.1 136.4 134.4 134.3 240 132.3 136.6 134.6 134.5 241 132.4 136.8 134.7 134.6 242 132.6 136.9 134.9 134.8 243 132.8 137.1 135.1 135.0 244 132.8 137.3 135.3 135.1 245 133.1 137.3 135.4 135.3 246 133.2 137.6 135.5 135.4 247 133.4 137.7 135.7 135.6 248 133.6 137.8 135.8 135.7 249 133.7 138.1 136.0 135.9 250 133.8 138.1 136.1 136.0 251 134.1 138.3 136.3 136.2 252 134.2 138.4 136.4 136.4 253 134.3 138.6 136.6 136.5

PAGE 146

146 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 254 134.4 138.7 136.8 136.6 255 134.6 138.9 136.9 136.8 256 134.8 139.0 137.1 137.0 257 135.0 139.2 137.3 137.2 258 135.1 139.3 137.3 137.2 259 135.3 139.4 137.6 137.4 260 135.4 139.6 137.7 137.6 261 135.5 139.7 137.8 137.7 262 135.6 139.9 138.1 137.9 263 135.7 140.0 138.1 137.9 264 135.9 140.2 138.2 138.1 265 136.1 140.3 138.4 138.3 266 136.2 140.5 138.5 138.4 267 136.4 140.6 138.7 138.6 268 136.4 140.7 138.9 138.7 269 136.7 140.9 139.0 138.9 270 136.8 140.9 139.1 138.9 271 136.9 141.2 139.3 139.1 272 137.1 141.3 139.4 139.2 273 137.2 141.4 139.6 139.4 274 137.3 141.6 139.7 139.5 275 137.4 141.7 139.9 139.7 276 137.7 141.8 140.0 139.8 277 137.7 141.9 140.1 139.9 278 137.8 142.1 140.3 140.1 279 138.1 142.2 140.4 140.2 280 138.1 142.3 140.6 140.3 281 138.3 142.5 140.7 140.5 282 138.4 142.6 140.9 140.6 283 138.5 142.8 140.9 140.8 284 138.7 142.9 141.2 140.9 285 138.9 143.0 141.3 141.1 286 139.0 143.1 141.4 141.2 287 139.1 143.3 141.6 141.3 288 139.3 143.4 141.7 141.5 289 139.4 143.6 141.8 141.6 290 139.6 143.7 141.9 141.8 291 139.7 143.8 142.1 141.9 292 139.9 143.9 142.3 142.0 293 140.1 144.1 142.5 142.2 294 140.2 144.2 142.6 142.3 295 140.4 144.4 142.7 142.5 296 140.5 144.5 142.9 142.6

PAGE 147

147 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 297 140.6 144.7 143.0 142.7 298 140.8 144.8 143.2 142.9 299 140.9 144.9 143.4 143.1 300 140.9 145.1 143.5 143.2 301 141.2 145.2 143.6 143.3 302 141.3 145.3 143.8 143.5 303 141.6 145.4 143.9 143.6 304 141.7 145.6 144.1 143.8 305 141.8 145.7 144.2 143.9 306 141.9 145.8 144.4 144.1 307 142.1 146.1 144.6 144.2 308 142.3 146.2 144.7 144.4 309 142.4 146.3 144.9 144.6 310 142.6 146.4 144.9 144.7 311 142.7 146.6 145.2 144.8 312 142.9 146.7 145.3 145.0 313 143.0 146.8 145.4 145.1 314 143.2 147.0 145.6 145.3 315 143.3 147.1 145.7 145.4 316 143.5 147.3 145.8 145.6 317 143.7 147.4 146.1 145.7 318 143.8 147.6 146.2 145.9 319 144.0 147.7 146.3 146.0 320 144.1 147.9 146.6 146.2 321 144.3 148.0 146.6 146.3 322 144.5 148.2 146.8 146.5 323 144.7 148.3 147.0 146.6 324 144.8 148.4 147.1 146.8 325 144.9 148.6 147.3 146.9 326 145.2 148.7 147.5 147.1 327 145.3 148.9 147.6 147.3 328 145.4 149.1 147.7 147.4 329 145.7 149.3 147.9 147.6 330 145.8 149.4 148.1 147.8 331 146.1 149.4 148.2 147.9 332 146.2 149.7 148.4 148.1 333 146.3 149.8 148.5 148.2 334 146.6 149.9 148.7 148.4 335 146.7 150.2 148.9 148.6 336 146.9 150.2 148.9 148.7 337 147.1 150.4 149.2 148.9 338 147.3 150.6 149.3 149.1 339 147.5 150.8 149.4 149.3

PAGE 148

148 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 340 147.6 150.9 149.7 149.4 341 147.9 151.1 149.8 149.6 342 148.0 151.3 150.1 149.8 343 148.2 151.5 150.2 149.9 344 148.4 151.6 150.2 150.1 345 148.6 151.8 150.4 150.3 346 148.8 152.0 150.7 150.5 347 148.9 152.1 150.8 150.6 348 149.2 152.3 151.1 150.8 349 149.4 152.4 151.2 151.0 350 149.6 152.6 151.3 151.2 351 149.8 152.8 151.5 151.4 352 149.9 153.0 151.7 151.6 353 150.2 153.1 151.9 151.7 354 150.2 153.3 152.1 151.9 355 150.4 153.4 152.2 152.0 356 150.7 153.6 152.4 152.2 357 150.8 153.8 152.6 152.4 358 151.1 153.8 152.8 152.6 359 151.3 153.9 153.0 152.8 360 151.5 154.2 153.1 152.9 361 151.6 154.4 153.3 153.1 362 151.9 154.6 153.4 153.3 363 152.1 154.7 153.7 153.5 364 152.2 154.9 153.8 153.6 365 152.4 155.1 153.9 153.8 366 152.6 155.2 154.2 154.0 367 152.8 155.4 154.3 154.2 368 153.0 155.6 154.6 154.4 369 153.2 155.8 154.7 154.6 370 153.4 156.0 154.8 154.7 371 153.7 156.2 155.1 155.0 372 153.8 156.4 155.2 155.1 373 153.9 156.5 155.4 155.3 374 154.2 156.7 155.5 155.4 375 154.3 156.9 155.7 155.6 376 154.6 157.1 155.9 155.8 377 154.7 157.2 156.1 156.0 378 154.9 157.5 156.3 156.2 379 155.1 157.6 156.5 156.4 380 155.3 157.8 156.6 156.6 381 155.5 157.9 156.9 156.8 382 155.7 158.2 157.0 157.0

PAGE 149

149 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 383 155.9 158.3 157.2 157.1 384 156.1 158.4 157.4 157.3 385 156.3 158.7 157.6 157.5 386 156.5 158.8 157.7 157.7 387 156.7 159.1 157.9 157.9 388 156.9 159.1 158.1 158.0 389 157.1 159.3 158.3 158.2 390 157.3 159.5 158.4 158.4 391 157.5 159.7 158.6 158.6 392 157.7 159.8 158.7 158.7 393 157.9 160.1 158.9 159.0 394 158.1 160.3 159.1 159.2 395 158.3 160.4 159.3 159.3 396 158.4 160.6 159.4 159.5 397 158.7 160.9 159.6 159.7 398 158.7 161.0 159.8 159.9 399 158.9 161.2 160.0 160.0 400 159.1 161.4 160.2 160.2 401 159.3 161.6 160.4 160.4 402 159.4 161.7 160.6 160.6 403 159.6 161.9 160.8 160.8 404 159.8 161.9 160.9 160.9 405 160.0 162.2 161.1 161.1 406 160.1 162.3 161.3 161.2 407 160.3 162.6 161.5 161.5 408 160.5 162.7 161.6 161.6 409 160.7 162.8 161.9 161.8 410 160.9 163.1 161.9 162.0 411 161.0 163.2 162.2 162.1 412 161.2 163.3 162.3 162.3 413 161.4 163.6 162.4 162.5 414 161.5 163.7 162.7 162.6 415 161.7 163.8 162.7 162.7 416 161.9 164.0 163.1 163.0 417 161.9 164.2 163.2 163.1 418 162.2 164.4 163.3 163.3 419 162.3 164.6 163.4 163.4 420 162.4 164.8 163.6 163.6 421 162.7 165.0 163.8 163.8 422 162.7 165.2 163.9 163.9 423 162.9 165.3 164.1 164.1 424 163.1 165.4 164.3 164.3 425 163.2 165.6 164.4 164.4

PAGE 150

150 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 426 163.4 165.8 164.6 164.6 427 163.6 165.9 164.8 164.7 428 163.7 166.1 165.0 164.9 429 163.9 166.2 165.2 165.1 430 164.1 166.3 165.4 165.3 431 164.2 166.6 165.5 165.4 432 164.3 166.7 165.7 165.6 433 164.6 166.8 165.8 165.7 434 164.7 166.9 165.9 165.9 435 164.9 167.2 166.1 166.0 436 165.1 167.2 166.3 166.2 437 165.2 167.3 166.4 166.3 438 165.4 167.6 166.6 166.5 439 165.5 167.6 166.8 166.6 440 165.7 167.8 166.8 166.8 441 165.8 167.9 166.9 166.9 442 165.9 168.1 167.2 167.1 443 166.1 168.2 167.2 167.2 444 166.2 168.4 167.3 167.3 445 166.3 168.5 167.6 167.5 446 166.4 168.7 167.6 167.6 447 166.7 168.8 167.8 167.8 448 166.8 169.0 167.9 167.9 449 166.8 169.1 168.0 168.0 450 167.1 169.2 168.2 168.1 451 167.2 169.4 168.3 168.3 452 167.2 169.5 168.5 168.4 453 167.3 169.7 168.6 168.5 454 167.6 169.8 168.7 168.7 455 167.6 170.0 168.9 168.8 456 167.7 170.1 169.0 168.9 457 167.9 170.3 169.1 169.1 458 168.0 170.4 169.3 169.2 459 168.1 170.4 169.4 169.3 460 168.3 170.7 169.5 169.5 461 168.4 170.8 169.7 169.6 462 168.5 170.8 169.8 169.7 463 168.6 170.9 169.9 169.8 464 168.7 171.2 170.0 170.0 465 168.9 171.2 170.2 170.1 466 169.0 171.3 170.3 170.2 467 169.1 171.6 170.4 170.4 468 169.2 171.6 170.4 170.4

PAGE 151

151 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 469 169.4 171.7 170.6 170.6 470 169.5 171.8 170.8 170.7 471 169.6 172.0 170.8 170.8 472 169.7 172.1 170.9 170.9 473 169.9 172.2 171.1 171.1 474 170.0 172.4 171.2 171.2 475 170.1 172.5 171.3 171.3 476 170.2 172.6 171.6 171.4 477 170.3 172.7 171.6 171.5 478 170.4 172.9 171.7 171.7 479 170.6 173.0 171.8 171.8 480 170.7 173.1 171.9 171.9 481 170.8 173.2 172.0 172.0 482 170.8 173.3 172.2 172.1 483 170.9 173.4 172.3 172.2 484 171.1 173.5 172.4 172.3 485 171.2 173.7 172.5 172.5 486 171.2 173.7 172.7 172.5 487 171.3 173.9 172.8 172.7 488 171.6 174.0 172.8 172.8 489 171.6 174.1 172.9 172.9 490 171.7 174.2 173.1 173.0 491 171.8 174.3 173.2 173.1 492 171.9 174.4 173.3 173.2 493 172.0 174.4 173.4 173.3 494 172.1 174.6 173.5 173.4 495 172.3 174.7 173.6 173.5 496 172.4 174.8 173.7 173.6 497 172.5 174.9 173.9 173.8 498 172.6 175.1 174.0 173.9 499 172.7 175.2 174.1 174.0 500 172.9 175.3 174.2 174.1 501 173.0 175.3 174.3 174.2 502 173.1 175.4 174.4 174.3 503 173.2 175.6 174.6 174.4 504 173.4 175.7 174.6 174.5 505 173.5 175.7 174.7 174.6 506 173.6 175.8 174.8 174.8 507 173.7 175.8 174.9 174.8 508 173.8 176.1 175.1 175.0 509 173.9 176.1 175.2 175.1 510 174.0 176.2 175.3 175.2 511 174.2 176.3 175.4 175.3

PAGE 152

152 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 512 174.3 176.4 175.4 175.4 513 174.4 176.5 175.7 175.5 514 174.4 176.6 175.7 175.6 515 174.7 176.7 175.8 175.7 516 174.8 176.9 176.1 175.9 517 174.8 177.0 176.1 176.0 518 174.9 177.1 176.2 176.1 519 175.1 177.2 176.4 176.2 520 175.3 177.3 176.5 176.4 521 175.3 177.4 176.6 176.4 522 175.4 177.5 176.7 176.6 523 175.6 177.7 176.9 176.7 524 175.7 177.8 176.9 176.8 525 175.8 177.9 177.1 176.9 526 176.1 178.0 177.2 177.1 527 176.1 178.1 177.4 177.2 528 176.3 178.2 177.4 177.3 529 176.4 178.3 177.6 177.4 530 176.5 178.5 177.7 177.6 531 176.6 178.6 177.9 177.7 532 176.8 178.7 178.0 177.8 533 176.9 178.8 178.2 177.9 534 177.0 178.9 178.3 178.1 535 177.2 179.1 178.4 178.2 536 177.3 179.2 178.5 178.3 537 177.4 179.3 178.7 178.4 538 177.6 179.6 178.8 178.6 539 177.8 179.7 178.9 178.8 540 177.9 179.8 179.1 178.9 541 178.0 179.9 179.2 179.0 542 178.2 180.1 179.3 179.2 543 178.4 180.2 179.6 179.4 544 178.5 180.3 179.7 179.5 545 178.7 180.4 179.8 179.6 546 178.9 180.6 179.9 179.8 547 178.9 180.7 180.1 179.9 548 179.2 180.9 180.2 180.1 549 179.3 181.0 180.3 180.2 550 179.6 181.1 180.5 180.4 551 179.7 181.3 180.6 180.5 552 179.8 181.5 180.8 180.7 553 179.9 181.6 180.9 180.8 554 180.1 181.8 181.1 181.0

PAGE 153

153 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 555 180.3 182.0 181.2 181.2 556 180.4 182.1 181.4 181.3 557 180.7 182.3 181.5 181.5 558 180.8 182.5 181.7 181.7 559 181.0 182.6 181.9 181.8 560 181.2 182.8 182.0 182.0 561 181.3 182.9 182.2 182.1 562 181.5 183.1 182.3 182.3 563 181.7 183.3 182.5 182.5 564 181.8 183.3 182.7 182.6 565 182.0 183.6 182.8 182.8 566 182.2 183.7 182.9 182.9 567 182.3 183.8 183.2 183.1 568 182.5 184.1 183.3 183.3 569 182.7 184.2 183.3 183.4 570 182.9 184.3 183.6 183.6 571 183.1 184.6 183.7 183.8 572 183.2 184.6 183.9 183.9 573 183.3 184.8 184.1 184.1 574 183.6 185.0 184.2 184.3 575 183.7 185.2 184.4 184.5 576 183.9 185.4 184.6 184.6 577 184.1 185.5 184.8 184.8 578 184.2 185.8 185.0 185.0 579 184.4 185.9 185.1 185.1 580 184.6 186.0 185.3 185.3 581 184.8 186.2 185.5 185.5 582 185.0 186.4 185.6 185.7 583 185.1 186.6 185.8 185.8 584 185.3 186.8 186.0 186.0 585 185.5 186.9 186.2 186.2 586 185.7 187.1 186.3 186.4 587 185.9 187.3 186.5 186.6 588 186.1 187.6 186.7 186.8 589 186.3 187.7 186.9 186.9 590 186.5 187.8 187.0 187.1 591 186.6 188.1 187.2 187.3 592 186.8 188.2 187.3 187.4 593 187.0 188.4 187.6 187.7 594 187.2 188.6 187.7 187.8 595 187.4 188.7 187.8 188.0 596 187.6 188.9 188.1 188.2 597 187.8 189.1 188.2 188.4

PAGE 154

154 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 598 187.8 189.3 188.4 188.5 599 188.2 189.4 188.6 188.7 600 188.3 189.6 188.7 188.9 601 188.4 189.8 188.9 189.1 602 188.6 190.0 189.1 189.2 603 188.8 190.2 189.3 189.5 604 188.9 190.3 189.5 189.6 605 189.1 190.5 189.6 189.7 606 189.3 190.7 189.8 189.9 607 189.5 191.0 190.0 190.2 608 189.6 191.2 190.2 190.3 609 189.8 191.4 190.3 190.5 610 190.0 191.6 190.5 190.7 611 190.2 191.8 190.8 190.9 612 190.4 191.9 191.1 191.1 613 190.6 192.2 191.2 191.3 614 190.7 192.3 191.4 191.5 615 191.0 192.6 191.6 191.7 616 191.2 192.7 191.7 191.9 617 191.3 192.9 191.9 192.1 618 191.4 193.1 192.1 192.2 619 191.6 193.3 192.2 192.4 620 191.8 193.5 192.4 192.6 621 191.9 193.7 192.6 192.8 622 192.1 193.8 192.7 192.9 623 192.2 194.0 192.9 193.1 624 192.4 194.3 193.1 193.3 625 192.6 194.4 193.3 193.4 626 192.7 194.6 193.4 193.6 627 192.9 194.8 193.6 193.8 628 193.0 195.0 193.7 193.9 629 193.2 195.2 194.0 194.1 630 193.3 195.3 194.1 194.2 631 193.5 195.6 194.3 194.4 632 193.7 195.7 194.5 194.6 633 193.8 195.9 194.6 194.8 634 194.0 196.1 194.8 194.9 635 194.1 196.3 195.0 195.1 636 194.3 196.4 195.1 195.3 637 194.5 196.6 195.3 195.4 638 194.6 196.7 195.4 195.6 639 194.7 197.1 195.6 195.8 640 194.9 197.1 195.8 195.9

PAGE 155

155 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 641 195.0 197.3 195.8 196.1 642 195.2 197.5 196.1 196.2 643 195.3 197.7 196.3 196.4 644 195.4 197.9 196.3 196.6 645 195.7 198.0 196.6 196.7 646 195.7 198.2 196.7 196.9 647 195.8 198.3 196.9 197.0 648 196.1 198.4 197.1 197.2 649 196.2 198.6 197.2 197.3 650 196.3 198.8 197.3 197.5 651 196.4 198.9 197.5 197.6 652 196.6 199.1 197.7 197.8 653 196.7 199.2 197.8 197.9 654 196.8 199.4 198.0 198.1 655 197.1 199.6 198.2 198.3 656 197.1 199.8 198.3 198.4 657 197.3 199.9 198.4 198.5 658 197.4 200.1 198.6 198.7 659 197.6 200.2 198.7 198.8 660 197.7 200.3 198.9 199.0 661 197.9 200.6 199.0 199.1 662 198.0 200.7 199.2 199.3 663 198.1 200.8 199.3 199.4 664 198.3 200.9 199.5 199.6 665 198.4 201.1 199.7 199.7 666 198.5 201.2 199.8 199.8 667 198.7 201.3 199.9 200.0 668 198.8 201.6 200.1 200.1 669 199.0 201.7 200.2 200.3 670 199.1 201.8 200.3 200.4 671 199.2 201.9 200.4 200.5 672 199.4 202.1 200.7 200.7 673 199.5 202.2 200.7 200.8 674 199.6 202.3 200.8 200.9 675 199.7 202.4 200.9 201.0 676 199.9 202.6 201.2 201.2 677 199.9 202.7 201.2 201.3 678 200.1 202.9 201.3 201.4 679 200.2 203.0 201.6 201.6 680 200.3 203.2 201.6 201.7 681 200.4 203.3 201.7 201.8 682 200.6 203.5 201.8 202.0 683 200.7 203.6 201.9 202.1

PAGE 156

156 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 684 200.8 203.8 202.1 202.2 685 200.9 203.8 202.2 202.3 686 201.1 203.9 202.3 202.4 687 201.1 204.1 202.4 202.5 688 201.2 204.2 202.6 202.7 689 201.3 204.3 202.7 202.8 690 201.4 204.4 202.8 202.9 691 201.6 204.6 202.9 203.0 692 201.6 204.7 203.1 203.1 693 201.7 204.8 203.3 203.3 694 201.8 204.9 203.4 203.4 695 201.9 204.9 203.5 203.5 696 202.0 205.1 203.6 203.5 697 202.1 205.2 203.7 203.6 698 202.2 205.3 203.9 203.8 699 202.3 205.4 203.9 203.9 700 202.4 205.6 203.9 204.0 701 202.5 205.6 204.1 204.1 702 202.6 205.7 204.3 204.2 703 202.7 205.7 204.4 204.3 704 202.8 205.8 204.4 204.4 705 202.9 205.9 204.6 204.5 706 203.1 206.0 204.7 204.6 707 203.2 206.1 204.8 204.7 708 203.3 206.2 204.8 204.8 709 203.4 206.3 204.9 204.9 710 203.5 206.4 205.2 205.0 711 203.6 206.5 205.2 205.1 712 203.7 206.5 205.2 205.1 713 203.9 206.7 205.4 205.4 714 203.9 206.8 205.6 205.4 715 204.1 206.9 205.6 205.5 716 204.2 207.0 205.7 205.6 717 204.3 207.1 205.8 205.7 718 204.4 207.2 205.8 205.8 719 204.4 207.3 205.9 205.9 720 204.6 207.3 206.1 206.0 721 204.7 207.5 206.2 206.1 722 204.8 207.6 206.2 206.2 723 204.8 207.7 206.4 206.3 724 205.1 207.8 206.5 206.4 725 205.2 208.0 206.6 206.6 726 205.2 208.1 206.7 206.7

PAGE 157

157 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 727 205.3 208.2 206.8 206.8 728 205.4 208.3 206.9 206.9 729 205.6 208.4 207.0 207.0 730 205.7 208.4 207.1 207.1 731 205.8 208.6 207.2 207.2 732 205.9 208.7 207.3 207.3 733 206.0 208.8 207.5 207.4 734 206.1 208.8 207.6 207.5 735 206.3 208.9 207.7 207.6 736 206.4 209.2 207.8 207.8 737 206.5 209.2 207.9 207.9 738 206.7 209.4 208.0 208.1 739 206.9 209.6 208.1 208.2 740 207.0 209.6 208.3 208.3 741 207.1 209.7 208.4 208.4 742 207.3 209.9 208.4 208.6 743 207.4 210.0 208.6 208.6 744 207.6 210.1 208.8 208.8 745 207.7 210.3 208.8 208.9 746 207.8 210.5 208.9 209.1 747 208.0 210.6 209.1 209.2 748 208.1 210.7 209.3 209.4 749 208.3 210.8 209.4 209.5 750 208.4 211.0 209.6 209.6 751 208.6 211.1 209.7 209.8 752 208.7 211.3 209.8 209.9 753 208.8 211.4 209.9 210.1 754 208.9 211.6 210.1 210.2 755 209.1 211.8 210.2 210.4 756 209.3 211.9 210.4 210.5 757 209.4 212.1 210.5 210.7 758 209.6 212.3 210.7 210.9 759 209.7 212.4 210.8 211.0 760 209.9 212.6 211.0 211.2 761 210.0 212.8 211.1 211.3 762 210.2 212.9 211.3 211.5 763 210.4 213.1 211.5 211.6 764 210.6 213.2 211.6 211.8 765 210.7 213.4 211.8 212.0 766 210.9 213.7 212.0 212.2 767 211.1 213.8 212.1 212.3 768 211.3 214.1 212.3 212.5 769 211.4 214.2 212.4 212.7

PAGE 158

158 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 770 211.6 214.3 212.7 212.9 771 211.8 214.5 212.8 213.0 772 211.9 214.7 212.9 213.2 773 212.1 214.9 213.2 213.4 774 212.4 215.1 213.3 213.6 775 212.4 215.4 213.4 213.8 776 212.7 215.6 213.7 214.0 777 212.8 215.8 213.9 214.2 778 213.1 216.0 214.1 214.4 779 213.2 216.2 214.3 214.6 780 213.4 216.4 214.5 214.8 781 213.7 216.6 214.7 215.0 782 213.9 216.8 214.9 215.2 783 214.1 217.1 215.1 215.4 784 214.2 217.3 215.3 215.6 785 214.4 217.4 215.6 215.8 786 214.7 217.7 215.7 216.0 787 214.9 217.8 216.0 216.2 788 215.1 218.1 216.2 216.5 789 215.4 218.3 216.3 216.7 790 215.7 218.6 216.4 216.9 791 215.9 218.8 216.8 217.2 792 216.1 219.0 216.9 217.3 793 216.3 219.3 217.2 217.6 794 216.6 219.5 217.3 217.8 795 216.8 219.8 217.6 218.0 796 216.9 220.0 217.7 218.2 797 217.2 220.3 218.1 218.5 798 217.4 220.5 218.2 218.7 799 217.7 220.8 218.4 219.0 800 217.9 220.9 218.6 219.2 801 218.1 221.3 218.9 219.4 802 218.3 221.4 219.1 219.6 803 218.6 221.7 219.5 219.9 804 218.8 222.1 219.7 220.2 805 219.1 222.2 220.0 220.4 806 219.5 222.5 220.2 220.7 807 219.8 222.8 220.5 221.0 808 220.1 223.1 220.7 221.3 809 220.3 223.4 220.9 221.5 810 220.6 223.7 221.3 221.9 811 220.9 224.0 221.4 222.1 812 221.1 224.3 221.7 222.4

PAGE 159

159 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 813 221.3 224.5 221.9 222.6 814 221.6 224.8 222.2 222.9 815 221.8 224.9 222.4 223.1 816 222.1 225.3 222.7 223.4 817 222.3 225.7 223.1 223.7 818 222.6 225.8 223.4 223.9 819 222.9 226.2 223.7 224.3 820 223.3 226.4 224.0 224.6 821 223.6 226.7 224.3 224.9 822 223.9 227.1 224.6 225.2 823 224.1 227.4 224.8 225.4 824 224.4 227.7 225.1 225.7 825 224.7 228.0 225.4 226.0 826 224.9 228.3 225.7 226.3 827 225.3 228.7 225.9 226.6 828 225.7 229.0 226.2 227.0 829 225.9 229.3 226.6 227.3 830 226.2 229.6 226.8 227.5 831 226.6 229.8 227.2 227.9 832 226.9 230.2 227.5 228.2 833 227.3 230.6 227.9 228.6 834 227.6 230.9 228.2 228.9 835 227.9 231.2 228.5 229.2 836 228.2 231.5 228.8 229.5 837 228.5 231.9 229.1 229.8 838 228.9 232.2 229.4 230.2 839 229.2 232.5 229.7 230.4 840 229.4 232.7 230.1 230.7 841 229.8 233.1 230.3 231.1 842 230.2 233.4 230.6 231.4 843 230.4 233.7 231.1 231.7 844 230.8 233.9 231.3 232.0 845 231.1 234.3 231.6 232.3 846 231.5 234.7 231.9 232.7 847 231.8 234.9 232.3 233.0 848 232.1 235.2 232.6 233.3 849 232.4 235.5 232.9 233.6 850 232.7 235.8 233.2 233.9 851 233.0 236.1 233.4 234.2 852 233.3 236.5 233.8 234.5 853 233.6 236.8 234.2 234.8 854 233.8 237.1 234.4 235.1 855 234.3 237.4 234.7 235.5

PAGE 160

160 Table D 1. Continued Time (sec) Bean Temp Bean Temp Bean Temp Average T (C) 856 234.6 237.7 235.1 235.8 857 234.9 237.9 235.4 236.1 858 235.2 238.3 235.7 236.4 859 235.4 238.7 236.1 236.7 860 235.8 238.9 236.4 237.1 861 236.1 239.2 236.8 237.4 862 236.4 239.5 237.1 237.6 863 236.6 239.8 237.4 237.9 864 236.9 240.1 237.7 238.2 865 237.3 237.9 237.6 866 237.6 238.3 237.9 867 237.9 238.7 238.3 868 238.2 239.1 238.6 869 238.4 239.2 238.8 870 238.8 239.5 239.2 871 239.1 239.8 239.5 872 239.4 240.1 239.8 873 239.7 239.7 874 239.9 239.9 875 240.3 240.3

PAGE 161

161 Table D 2 L*, a*, and b* values of coffee samples roasted under non isothermal conditions in an Ambex Roaster at various temperatures (Chapter 3) Roasting T (C) L* a* b* 20 54.47 1.8 17.03 59.89 0.09 17.12 63.53 1.09 16.67 120 58.9 0.1 15.97 63.49 1.26 15.69 63.31 1.49 15.39 140 62.95 0.21 16.35 63.37 0.11 16.74 64.18 0.96 15.34 160 59.46 6.54 26.56 62.24 5.9 26.45 61.69 4.71 24.3 170 54.22 13.07 31.43 56.59 12.25 31.66 55.89 11.67 30.26 180 45.5 17.9 32.86 46.66 17.74 33.33 46.03 16.47 30.62 190 39.22 19.21 32.93 39.07 18.96 32.48 38.82 17.65 29.93 200 30.89 19.87 29.43 31 19.58 29.31 31.36 17.5 26.85 210 20.44 19.63 22.02 20.82 18.97 21.75 22.7 16.84 21.03 220 14.1 15.77 12.09 14.54 15.74 12.51 16.48 14.35 12.97 230 9.6 11.69 7.78 10.78 12.17 9.02 10.97 8.13 6.38 240 5.46 2.29 0.09 5.91 2.39 0.1 7.98 2.37 0.43

PAGE 162

162 LIST OF REFERENCES Adams A, Borrelli RC, Fogliano V, De Kimpe N. 2005. Thermal degradation studies of food melanoidins. J. Agric. Food Chem. 53(10):4136 42. Anzueto F, Baumann TW, Graziosi G, Piccin CR, Sondahl MR, van der Vossen HAM. 2005. The plant. In: Illy A, Viani R, editors. Espresso coffee The science of quality 2 nd ed. San Diego, CA: Elsevier Academic Press. p 21 86. Arya M, Rao LJM. 2007. An impression of coffee carbohydrates. Crit. Rev. Food Sci. Nutr. 47(1):51 67. Baggenstoss J, Thomann D, Perren R, Escher F. 2010. Aroma recovery from roasted coffee by wet grindi ng. J. Food Sci. 75(9):C697 C702. Baggenstoss J, Poisson L, Luethi R, Perren R, Escher F. 2007. Influence of water quench cooling on degassing and aroma stability of roasted coffee. J. Agric. Food Chem. 55(16):6685 91. Baggenstoss J, Poisson L, Kaegi R, Pe rren R, Eschert F. 2008a. Roasting and aroma formation: effect of initial moisture content and steam treatment. J. Agric. Food Chem. 56(14):5847 51. Baggenstoss J, Poisson L, Kaegi R, Perren R, Escher F. 2008b. Coffee roasting and aroma formation: applicat ion of different time temperature conditions. J. Agric. Food Chem. 56(14):5836 46. Balaban MO. 2008. Quantifying nonhomogeneous colors in agricultural materials part I: Method development. J. Food Sci. 73(9):S431 S7. Balaban MO, Odabasi AZ. 2006. Measuring color with machine vision. Food Technol. 60(12):32 6. Balzer HH. 2001. Chemistry I: non volatile compounds (acids in coffee). In: Clarke RJ, Vitzhum OG, editors. Coffee: recent developments 1 st ed. Malden, MA: Blackwell Science Ltd. p 18 32. Bee S, Brand o CHJ, Brumen G, Carvalhaes N, Kolling Speer I, Speer K, Suggi Liverani F, Teixeira AA, Teixeira R, Thomaziello RA, Viani R, Vitzhum OG. 2005. The raw bean. In: Illy A, Viani R, editors. Espresso coffee the science of quality 2 nd ed. San Diego, CA: Else vier Academic Press. p 87 178. Bicchi CP, Binello AE, Pellegrino GM, Vanni AC. 1995. Characterization of green and roasted coffees through the chlorogenic acid fraction by HPLC UV and principal component analysis. J. Agric. Food Chem. 43(6):1549 55.

PAGE 163

163 Bonnlander B, Eggers R, Engelhardt UH, Maier HG. 2005. Roasting. In: Illy A, Viani R, editors. Espresso coffee the science of quality 2 nd ed. San Diego, CA: Elsevier Academic Press. p 179 214. Borrelli RC, Visconti A, Mennella C, Anese M, Fogliano V. 20 02. Chemical characterization and antioxidant properties of coffee melanoidins. J. Agric. Food Chem. 50(22):6527 33. Bradbury AGW. 2001. Chemistry I: non volatile compounds (carbohydrates). In: Clarke RJ, Vitzhum OG, editors. Coffee: recent developments 1 st ed. Malden, MA: Blackwell Science Ltd. p 1 17. Brosnan T, Sun DW. 2002. Inspection and grading of agricultural and food products by computer vision systems a review. Comput. Electron. Agric. 36(2 3):193 213. Broyart B, Trystram G, Duquenoy A. 1998. Pr edicting colour kinetics during cracker baking. J. Food Eng. 35(3):351 68. Buffo RA, Cardelli Freire C. 2004. Coffee flavour: an overview. Flavour Frag. J. 19(2):99 104. Buttery RG, Guadagni DG, Ling LC, Seifert RM, Lipton W. 1976. Additional volatile comp onents of cabbage, broccoli, and cauliflower. J. Agric. Food Chem. 24(4):829 32. Clifford MN, Jarvis T. 1988. The chlorogenic acids content of green robusta coffee beans as a possible index of geographic origin. Food Chem. 29(4):291 8. Clifford MN, William s T, Bridson D. 1989. Chlorogenic acids and caffeine as possible taxonomic criteria in Coffea and Psilanthus. Phytochemistry 28(3):829 38. Czerny M, Mayer F, Grosch W. 1999. Sensory study on the character impact odorants of roasted arabica coffee. J. Agric Food Chem. 47(2):695 9. Davids K. 2003. Home coffee roasting 2 nd ed. New York: St. Martin's Press. 245 p. Demir AD, Celayeta JMF, Cronin K, Abodayeh K. 2002. Modelling of the kinetics of colour change in hazelnuts during air roasting. J. Food Eng. 55(4) :283 92. Eggers R, Pietsch A. 2001. Technology I: roasting. In: Clarke RJ, Vitzhum OG, editors. Coffee: recent developments 1 st ed. Malden, MA: Blackwell Science Ltd. p 90 107. FAOSTAT: Coffee production [internet]. Rome, Italy: Food and Agriculture Organ ization Statistics Division; 2011 [Accessed 2011 Feb 7]. Available from: http://faostat.fao.org/

PAGE 164

164 Farah A, De Paulis T, Trugo LC, Martin PR. 2005. Effect of roasting on the formation of chlorogenic acid lactones in coffee. J. Agric. Food Chem. 53(5):1505 13. Farah A, Monteiro MC, Calado V, Franca AS, Trugo LC. 2006a. Correlation between cup quality and chemical attributes of Brazilian coffee. Food Chem. 98(2):373 80. Farah A, De Paulis T, Moreira DP, Trugo LC, Martin PR. 2006b. Chlorogenic acids and lactones in regular and water decaffeinated arabica coffees. J. Agric. Food Chem. 54(2):374 81. Feldman JR, Ryder WS, Kung JT. 1969. Importance of nonvaltile compounds to flavor of coffee. J. Agric. Food Chem. 17(4):733 &. Flament I. 2001. Coffee flavor chemistry 1 st ed. New York: John Wiley and Sons, Ltd. 424 p. Franca AS, Oliveira LS, Oliveira RCS, Agresti PCM, Augusti R. 2009. A preliminary evaluation of the effect of processing temperature on coffee roasting degree ass essment. J. Food Eng. 92(3):345 52. Fujioka K, Shibamoto T. 2008. Chlorogenic acid and caffeine contents in various commercial brewed coffees. Food Chem. 106(1):217 21. Gniechwitz D, Reichardt N, Ralph J, Blaut M, Steinhart H, Bunzel M. 2008. Isolation and characterisation of a coffee melanoidin fraction. J. Sci. Food Agric. 88(12):2153 60. Gokmen V, Senyuva HZ. 2006. Study of colour and acrylamide formation in coffee, wheat flour and potato chips during heating. Food Chem. 99(2):238 43. Grosch W. 1998. Fla vour of coffee. A review. Food / Nahrung 42(06):344 50. Grosch W. 2001. Chemistry III: volatile compounds. In: Clarke RJ, Vitzhum OG, editors. Coffee: recent developments 1 st ed. Malden, MA: Blackwell Science Ltd. p 68 89. Grosch W, Mayer F. 2000. Release of odorants from roasted coffee. In: Roberts DD, Taylor AJ, editors. Flavor release Washington, DC: American Chemical Society. p 430 8. Guth H, Grosch W. 1994. Identification of the Character Impact Odorants of Stewed Beef Juice by Instrumental Analyses and Sensory Studies. J. Agric. Food Chem. 42(12):2862 6. Hashim L, Chaveron H. 1995. Use of methylpyrazine ratios to monitor the coffee roasting. Food Res. Int. 28(6):619 23.

PAGE 165

165 Hernandez JA, Heyd B, Trystram G. 2008a. On line assessment of brightness and surface kinetics during coffee roasting. J. Food Eng. 87(3):314 22. Hernandez JA, Heyd B, Trystram G. 2008b. Prediction of brightness and surface area kinetics during coffee roasting. J. Food Eng. 89(2):156 63. Hernandez JA, Heyd B, Irles C, Valdovinos B, Trystram G. 2007. Analysis of the heat and mass transfer during coffee batch roasting. J. Food Eng. 78(4):1141 8. Hill CG, Griegerblock RA. 1980. Kinetic data generation, interpretation, and use. Food Technol. 34(2):56 66. Holscher W, Steinhart H. 1992. Formation pathways for primary roasted coffee aroma compounds. In: Parliment TH, Morello MJ, McGorrin RJ, editors. Thermally generated flavors: maillard, microwave, and extrusion processes Washington, DC: American Chemical Society. p 206 17. Holscher W, V itzthum OG, Steinhart H. 1992. Prenyl alcohol source for odorants in roasted coffee. J. Agric. Food Chem. 40(4):655 8. Homma S. 2001. Chemistry II: non volatile compounds, part II. In: Clarke RJ, vitzhum OG, editors. Coffee: recent developments 1 st ed. Malden, MA: Blackwell Science Ltd. p 50 67. Houessou JK, Maloug S, Leveque AS, Delteil C, Heyd B, Camel V. 2007. Effect of roasting conditions on the polycyclic aromatic hydrocarbon content in ground Arabica coffee and coffee brew. J. Agric. Food Chem. 55( 23):9719 26. Hutchings JB. 1999. Food color and appearance 2nd ed. Gaithersburg, MD: Aspen Publishers, Inc. 610 p. Ibarz A, Pagan J, Garza S. 1999. Kinetic models for colour changes in pear puree during heating at relatively high temperatures. J. Food Eng 39(4):415 22. ICO: Historical data [Internet]. International Coffee Organization; 2010a [Accessed 2011 Feb 7]. Available from: http://www.ico.org/ ICO: Story of coffee [Internet]. London, UK: International Coffee Organization; 2010b [Accessed 2011 Feb 7]. Available from: http://www.ico.org/ Kahyaoglu T, Kaya S. 2006. Modeling of moisture, color and texture changes in sesame seeds during the conventional roasting. J. Food Eng. 75 (2):167 77. Kim S, Park JB, Hwang IK. 2002. Quality attributes of various varieties of Korean red pepper powders (Capsicum annuum L.) and color stability during sunlight exposure. J. Food Sci. 67(8):2957 61.

PAGE 166

166 Kleinwachter M, Selmar D. 2010. Influence of dry ing on the content of sugars in wet processed green arabica coffees. Food Chem. 119(2):500 4. Krokida MK, Oreopoulou V, Maroulis ZB, Marinos Kouris D. 2001. Colour changes during deep fat frying. J. Food Eng. 48(3):219 25. Labuza TP, Riboh D. 1982. Theory and application of arrhenius kinetics to the prediction of nutrient losses in foods. Food Technol. 36(10):66 74. Lau MH, Tang J, Swanson BG. 2000. Kinetics of textural and color changes in green asparagus during thermal treatments. J. Food Eng. 45(4):231 6 Lee KG, Shibamoto T. 2002. Analysis of volatile components isolated from Hawaiian green coffee beans (Coffea arabica L.). Flavour Frag. J. 17(5):349 51. Little AC, Mackinney G. 1956. On the color of coffee. Food Technol. 10(10):503 6. Little AC, Chichest er CO, Mackinney G. 1958. On the color of coffee. II. Food Technol. 12(10):505 7. Little AC, Chichester CO, Mackinney G. 1959. On the color of coffee. III. Effect of roasting conditions on flavor development for a given color role of initial moisture lev el on roasting characteristics of green coffee beans. Food Technol. 13(12):684 8. Lockhart EE. 1960. Roasted coffee color measurement and classification. Food Technol. 14(11):597 600. Lozano JE, Ibarz A. 1997. Colour changes in concentrated fruit pulp du ring heating at high temperatures. J. Food Eng. 31(3):365 73. Lukac H, Amrein TM, Perren R, Conde Petit B, Amado R, Escher F. 2007. Influence of roasting conditions on the acrylamide content and the color of roasted almonds. J. Food Sci. 72(1):C33 C8. Luzu riaga DA, Balaban MO, Yeralan S. 1997. Analysis of visual quality attributes of white shrimp by machine vision. J. Food Sci. 62(1):113 &. MacDougall DB. 1988. Color vision and appearance measurement. In: Piggott JR, editor. Sensory analysis of foods Essex UK: Elsevier Science Publishers Ltd. p 103 30. Mahattanatawee K, Ruiz Perez Cacho MP, Rattanapanone N, Rouseff R. 2007. Aroma active sulphur volatiles in arabica coffee and degree of roasting effects. In: Schieberle P, Hofmann T, editors. Recent highligh ts flavor chem. biol., [Proc. Wartburg Symp.] Wartburg, Germany. p 111 6.

PAGE 167

167 Marin K, Pozrl T, Zlatic E, Plestenjak A. 2008. A new aroma index to determine the aroma quality of roasted and ground coffee during storage. Food Technol. Biotechnol. 46(4):442 7. Mayer F, Czerny M, Grosch W. 1999. Influence of provenance and roast degree on the composition of potent odorants in Arabica coffees. Eur. Food Res. Technol. 209(3):242 50. Mayer F, Czerny M, Grosch W. 2000. Sensory study of the character impact aroma comp ounds of a coffee beverage. Eur. Food Res. Technol. 211(4):272 6. Merritt MC, Proctor BE. 1959. Effect of temperature during the roasting cycle on selected components of different types of whole bean coffee. J. Food Sci. 24(6):672 80. Mills E, Borg N. 1999 Trends in recommended illuminance levels: An international comparison. J. Illum. Eng. Soc. 28(1):155 63. Montavon P, Mauron AF, Duruz E. 2003. Changes in green coffee protein profiles during roasting. J. Agric. Food Chem. 51(8):2335 43. Murkovic M. 2004. Acrylamide in Austrian foods. J. Biochem. Biophys. Methods 61(1 2):161 7. Murkovic M, Derler K. 2006. Analysis of amino acids and carbohydrates in green coffee. J. Biochem. Biophys. Methods 69(1 2):25 32. Mwithiga G, Jindal VK. 2003. Physical changes duri ng coffee roasting in rotary conduction type heating units. J. Food Process Eng. 26(6):543 58. Nagaraju V, Bhattacharya S. 2010. Roasting green coffee beans using spouted bed roaster: changes in physical characteristics. J. Food Sci. Technol. 47(6):674 7. Nagaraju VD, Murthy CT, Ramalakshmi K, Srinivasa Rao PN. 1997. Studies on roasting of coffee beans in a spouted bed. J. Food Eng. 31(2):263 70. Ortola MD, Londono L, Gutierrez CL, Chiralt A. 1998. Influence of roasting temperature on physicochemical proper ties of different coffees. Food Sci. Technol. Int. 4(1):59 66. Ozdemir M, Devres O. 2000. Analysis of color development during roasting of hazelnuts using response surface methodology. J. Food Eng. 45(1):17 24. Parliment TH. 2000. An overview of coffee roa sting. In: Parliment TH, Ho C T, Schieberle P, editors. Caffeinated beverages: health benefits, physiological effects, and chemistry Washington, DC: American Chemical Society. p 188 201.

PAGE 168

168 Parliment TH, Stahl HD. 1994. Generation of furfuryl mercaptan in cy steine pentose model systems in relation to roasted coffee. In: Mussinan C, Keelan ME, editors. Sulfur compounds in foods. Washington, DC: American Chemical Society. p 160 70. Parliment TH, Stahl HD. 1995. What makes that coffee smell so good. Chem. Tech. 25(8):38 47. Parras P, Martinez Tome M, Jimenez AM, Murcia MA. 2007. Antioxidant capacity of coffees of several origins brewed following three different procedures. Food Chem. 102(3):582 92. Pedreschi F, Mery D, Marique T. 2008. Quality Evaluation of Fruit and Vegetables Grading of Potatoes. In: Sun D W, editor. Computer vision technology for food quality evaluation Burlington, MA: Elsevier Inc. p 583. Pendergrast M. 2009. Coffee: second to oil? Tea Coffee Trade J 181(4):38 41. Pittia P, Dalla Rosa M, Le rici CR. 2001. Textural changes of coffee beans as affected by roasting conditions. Lebensmittel Wissenschaft und Technologie 34(3):168 75. Poisson L, Schmalzried F, Davidek T, Blank I, Kerler J. 2009. Study on the role of precursors in coffee flavor forma tion using in bean experiments. J. Agric. Food Chem. 57(21):9923 31. Pugash M. 1995. The roast color classification system: technology designed to advance the art & science of coffee roasting. Tea Coffee Trade J 167(7):66 74. Purdon MP, McCamey DA. 1987. U se of a 5 caffeoylquinic acid/caffeine ratio to monitor the coffee roasting process. J. Food Sci. 52(6):1680 3. Raemy A, Lambelet P. 1982. A calorimetric study of self heating in coffee and chicory. Int. J. Food Sci. Tech. 17(4):451 60. Reichstein T, Staud inger H, inventors; Int. Nahr. Genussm. AG, assignee. 1926. Improvements in a method for isolating the aromatic principle contained in roasted coffee. British patent 246,454. Rothe M. 1976. Aroma values a useful concept? Food / Nahrung 20(3):259 66. Sche nker S. 2000. Investigations on the hot air roasting of coffee beans. [DPhil dissertation]. Zurich: Swiss Federal Institute of Technology. 174 p. Available from: http://e collection.ethbib.ethz.ch/eserv/eth:23461/eth 23461 02.pdf#search=%22coffee%20roasting%22 Schenker S, Handschin S, Frey B, Perren R, Escher F. 2000. Pore structure of coffee beans affected by roasting conditions. J. Food Sci. 65(3):452 7.

PAGE 169

169 Schenker S, Heinemann C, Huber M, Pompizzi R, Perren R, Escher F. 2002. Impact of roasting conditions on the formation of aroma compounds in coffee beans. J. Food Sci. 67(1):60 6. Schneider R, Kotseridis Y, Ray J L, Augier C, Baumes R. 2003. Quantitative determination of sulfur containing wine odorants at sub parts per billion levels. 2. Development and application of a stable isotope dilution assay. J. Agric. Food Chem. 51(11):3243 8. Semmelroch P, Laskawy G, Blank I, Grosch W. 1995. De termination of potent odourants in roasted coffee by stable isotope dilution assays. Flavour Frag. J. 10(1):1 7. Shwartzberg HG. 2002. Modeling bean heating during batch roasting of coffee beans. In: Welti Chanes J, Barbosa Canovas GV, Aguilera JM, editors Engineering and food for the 21 st century Boca Raton, FL: CRC Press. p 871 90. Sivetz M. 1963. Coffee processing technology Westport, CT: Avi Pub Co. 379 p. Sivetz M, editor. Growth in use of automated fluid bed roasting of coffee beans. Proceedings of the 14 th ASIC Colloquium; 1991; San Francisco. Paris: ASIC; 1991. Sivetz M, Derousier NW. 1979. Coffee technology Westport, CT: AVI Pub Co. 716 p. Storey ML, Forshee RA, Anderson PA. 2006. Beverage consumption in the US population. J. Am. Diet. Assoc. 10 6(12):1992 2000. Summa CA, de la Calle B, Brohee M, Stadler RH, Anklam E. 2007. Impact of the roasting degree of coffee on the in vitro radical scavenging capacity and content of acrylamide. Food Sci. Technol. 40(10):1849 54. Taoukis PS, Labuza TP, Saguy I S. 1997. Kinetics of food deterioration and shelf life prediction. In: Valentas KJ, Rotstein E, Singh RP, editors. Handbook of food engineering practice Boca Raton, FL: CRC Press. p 361 402. Tressl R. 1989. Formation of flavor components in roasted coffee In: Parliment TH, McGorrin RJ, Ho CT, editors. Thermal generation of aromas Washington, DC: American Chemical Society. p 285 301. Trugo LC, Macrae R. 1984. A study of the effect of roasting on the cholorogenic acid composition of coffee using HPLC. Food Chem. 15(1):219 27. van Boekel MAJS. 2009. Kinetic modeling of reactions in foods 1 st ed. Boca Raton, FL: CRC Press. 767 p. Wasserman GS, Bradbury A, Cruz T, Penson S. 2000. Coffee Hoboken, NJ: John Wiley & Sons, Inc. 22 p.

PAGE 170

170 Yagiz Y, Balaban MO, Kristinsson HG, Welt BA, Marshall MR. 2009. Comparison of Minolta colorimeter and machine vision system in measuring colour of irradiated Atlantic salmon. J. Sci. Food Agric. 89(4):728 30.

PAGE 171

171 BIOGRAPHICAL SKETCH Alberto M. C. Azeredo was born in Brazil. He earned his B.S. degree in Food Engineering from University of Vicosa Brazil in 1995, and his M.S. degree from the same institution in 2000. He worked at Brahma Ambev during 1996 and 1997, at Dannon Company Inc. from 1999 to 2003, and at Fuchs Gewrze from 2003 to 2005. Alberto came to the U S in 2005 to work as a research assistant at the Food Science and Human Nutrition department of the University of Florida. He became a graduate student in 2006 and finished his PhD degree in Food Science in 2011, with a minor in Packaging Sciences. He has a strong interest in food process develop ment and in teaching, and intend s to pursue a career in the industry.